diff --git a/.github/workflows/docs.yml b/.github/workflows/docs.yml index 4e469df1730d..d2219e6c7ed2 100644 --- a/.github/workflows/docs.yml +++ b/.github/workflows/docs.yml @@ -35,7 +35,6 @@ jobs: { ref: "v0.4.0.dev0", dest-dir: "0.4.0.dev0" }, { ref: "v0.4.0.dev1", dest-dir: "0.4.0.dev1" }, { ref: "v0.4.0.dev2", dest-dir: "0.4.0.dev2" }, - { ref: "v0.4.0.dev3", dest-dir: "0.4.0.dev3" }, ] steps: - name: Checkout diff --git a/README.md b/README.md index 049612da46be..43bdd263d310 100644 --- a/README.md +++ b/README.md @@ -101,7 +101,7 @@ We look forward to your contributions! First install the packages: ```bash -pip install 'autogen-agentchat==0.4.0.dev3' 'autogen-ext[docker]==0.4.0.dev3' +pip install autogen-agentchat==0.4.0.dev2 autogen-ext==0.4.0.dev2 ``` The following code uses code execution, you need to have [Docker installed](https://docs.docker.com/engine/install/) @@ -109,11 +109,17 @@ and running on your machine. ```python import asyncio +import logging +from autogen_agentchat import EVENT_LOGGER_NAME +from autogen_agentchat.agents import CodeExecutorAgent, CodingAssistantAgent +from autogen_agentchat.logging import ConsoleLogHandler +from autogen_agentchat.teams import RoundRobinGroupChat, StopMessageTermination from autogen_ext.code_executor.docker_executor import DockerCommandLineCodeExecutor from autogen_ext.models import OpenAIChatCompletionClient -from autogen_agentchat.agents import CodeExecutorAgent, CodingAssistantAgent -from autogen_agentchat.teams import RoundRobinGroupChat -from autogen_agentchat.task import TextMentionTermination + +logger = logging.getLogger(EVENT_LOGGER_NAME) +logger.addHandler(ConsoleLogHandler()) +logger.setLevel(logging.INFO) async def main() -> None: async with DockerCommandLineCodeExecutor(work_dir="coding") as code_executor: @@ -121,13 +127,11 @@ async def main() -> None: coding_assistant_agent = CodingAssistantAgent( "coding_assistant", model_client=OpenAIChatCompletionClient(model="gpt-4o", api_key="YOUR_API_KEY") ) - termination = TextMentionTermination("TERMINATE") - group_chat = RoundRobinGroupChat([coding_assistant_agent, code_executor_agent], termination_condition=termination) - stream = group_chat.run_stream( - "Create a plot of NVDIA and TSLA stock returns YTD from 2024-01-01 and save it to 'nvidia_tesla_2024_ytd.png'." + group_chat = RoundRobinGroupChat([coding_assistant_agent, code_executor_agent]) + result = await group_chat.run( + task="Create a plot of NVDIA and TSLA stock returns YTD from 2024-01-01 and save it to 'nvidia_tesla_2024_ytd.png'.", + termination_condition=StopMessageTermination(), ) - async for message in stream: - print(message) asyncio.run(main()) ``` diff --git a/docs/switcher.json b/docs/switcher.json index 8bde73629168..364394a4f3f7 100644 --- a/docs/switcher.json +++ b/docs/switcher.json @@ -21,12 +21,7 @@ { "name": "0.4.0.dev2", "version": "0.4.0.dev2", - "url": "/autogen/0.4.0.dev2/" - }, - { - "name": "0.4.0.dev3", - "version": "0.4.0.dev3", - "url": "/autogen/0.4.0.dev3/", + "url": "/autogen/0.4.0.dev2/", "preferred": true } ] diff --git a/dotnet/.gitignore b/dotnet/.gitignore index 2fc32d9ac7e4..25f613c7945f 100644 --- a/dotnet/.gitignore +++ b/dotnet/.gitignore @@ -37,6 +37,9 @@ bld/ # vs code cache .vscode/ +# Properties +Properties/ + artifacts/ output/ @@ -53,6 +56,8 @@ bld/ [Ll]og/ [Ll]ogs/ +appsettings.json + # Visual Studio 2015/2017 cache/options directory .vs/ # Uncomment if you have tasks that create the project's static files in wwwroot diff --git a/dotnet/AutoGen.sln b/dotnet/AutoGen.sln index 4f82713b5adb..291cb484649d 100644 --- a/dotnet/AutoGen.sln +++ b/dotnet/AutoGen.sln @@ -68,13 +68,6 @@ EndProject Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "Solution Items", "Solution Items", "{243E768F-EA7D-4AF1-B625-0398440BB1AB}" ProjectSection(SolutionItems) = preProject .editorconfig = .editorconfig - .gitattributes = .gitattributes - .gitignore = .gitignore - Directory.Build.props = Directory.Build.props - Directory.Build.targets = Directory.Build.targets - Directory.Packages.props = Directory.Packages.props - global.json = global.json - NuGet.config = NuGet.config spelling.dic = spelling.dic EndProjectSection EndProject @@ -130,7 +123,7 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "HelloAgent", "samples\Hello EndProject Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "AIModelClientHostingExtensions", "src\Microsoft.AutoGen\Extensions\AIModelClientHostingExtensions\AIModelClientHostingExtensions.csproj", "{97550E87-48C6-4EBF-85E1-413ABAE9DBFD}" EndProject -Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Microsoft.AutoGen.Agents.Tests", "Microsoft.AutoGen.Agents.Tests\Microsoft.AutoGen.Agents.Tests.csproj", "{CF4C92BD-28AE-4B8F-B173-601004AEC9BF}" +Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Microsoft.AutoGen.Agents.Tests", "Microsoft.AutoGen.Agents.Tests\Microsoft.AutoGen.Agents.Tests.csproj", "{CF4C92BD-28AE-4B8F-B173-601004AEC9BF}" EndProject Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "sample", "sample", "{686480D7-8FEC-4ED3-9C5D-CEBE1057A7ED}" EndProject diff --git a/dotnet/Directory.Build.props b/dotnet/Directory.Build.props index 1e84f78232ad..bb78c84d14f4 100644 --- a/dotnet/Directory.Build.props +++ b/dotnet/Directory.Build.props @@ -13,7 +13,6 @@ CS1998;CS1591;CS8002; SKEXP0001;SKEXP0010;SKEXP0020 $(NoWarn);$(CSNoWarn);$(SKEXPNoWarn);NU5104 - true true false @@ -33,6 +32,10 @@ $(NoWarn);CA1829 + + + + diff --git a/dotnet/Directory.Packages.props b/dotnet/Directory.Packages.props index 75580bba007a..dd9df7161047 100644 --- a/dotnet/Directory.Packages.props +++ b/dotnet/Directory.Packages.props @@ -3,7 +3,7 @@ true 1.22.0 1.22.0-alpha - 9.0.0-preview.9.24525.1 + 9.0.0-preview.9.24507.7 @@ -111,5 +111,6 @@ + - + \ No newline at end of file diff --git a/dotnet/samples/AutoGen.BasicSamples/AutoGen.BasicSample.csproj b/dotnet/samples/AutoGen.BasicSamples/AutoGen.BasicSample.csproj index 9e510ffa2f1a..460c95f3743a 100644 --- a/dotnet/samples/AutoGen.BasicSamples/AutoGen.BasicSample.csproj +++ b/dotnet/samples/AutoGen.BasicSamples/AutoGen.BasicSample.csproj @@ -15,6 +15,5 @@ - diff --git a/dotnet/samples/AutoGen.BasicSamples/Example03_Agent_FunctionCall.cs b/dotnet/samples/AutoGen.BasicSamples/Example03_Agent_FunctionCall.cs index c1087080350d..ca05a42ca367 100644 --- a/dotnet/samples/AutoGen.BasicSamples/Example03_Agent_FunctionCall.cs +++ b/dotnet/samples/AutoGen.BasicSamples/Example03_Agent_FunctionCall.cs @@ -6,7 +6,6 @@ using AutoGen.OpenAI; using AutoGen.OpenAI.Extension; using FluentAssertions; -using Microsoft.Extensions.AI; /// /// This example shows how to add type-safe function call to an agent. @@ -38,20 +37,13 @@ public async Task ConcatString(string[] strings) /// /// price, should be an integer /// tax rate, should be in range (0, 1) - [Function] + [FunctionAttribute] public async Task CalculateTax(int price, float taxRate) { return $"tax is {price * taxRate}"; } - /// - /// This example shows how to add type-safe function call using AutoGen.SourceGenerator. - /// The SourceGenerator will automatically generate FunctionDefinition and FunctionCallWrapper during compiling time. - /// - /// For adding type-safe function call from M.E.A.I tools, please refer to . - /// - /// - public static async Task ToolCallWithSourceGenerator() + public static async Task RunAsync() { var instance = new Example03_Agent_FunctionCall(); var gpt4o = LLMConfiguration.GetOpenAIGPT4o_mini(); @@ -109,60 +101,4 @@ public static async Task ToolCallWithSourceGenerator() // send aggregate message back to llm to get the final result var finalResult = await agent.SendAsync(calculateTaxes); } - - /// - /// This example shows how to add type-safe function call from M.E.A.I tools. - /// - /// For adding type-safe function call from source generator, please refer to . - /// - public static async Task ToolCallWithMEAITools() - { - var gpt4o = LLMConfiguration.GetOpenAIGPT4o_mini(); - var instance = new Example03_Agent_FunctionCall(); - - AIFunction[] tools = [ - AIFunctionFactory.Create(instance.UpperCase), - AIFunctionFactory.Create(instance.ConcatString), - AIFunctionFactory.Create(instance.CalculateTax), - ]; - - var toolCallMiddleware = new FunctionCallMiddleware(tools); - - var agent = new OpenAIChatAgent( - chatClient: gpt4o, - name: "agent", - systemMessage: "You are a helpful AI assistant") - .RegisterMessageConnector() - .RegisterStreamingMiddleware(toolCallMiddleware) - .RegisterPrintMessage(); - - // talk to the assistant agent - var upperCase = await agent.SendAsync("convert to upper case: hello world"); - upperCase.GetContent()?.Should().Be("HELLO WORLD"); - upperCase.Should().BeOfType(); - upperCase.GetToolCalls().Should().HaveCount(1); - upperCase.GetToolCalls().First().FunctionName.Should().Be(nameof(UpperCase)); - - var concatString = await agent.SendAsync("concatenate strings: a, b, c, d, e"); - concatString.GetContent()?.Should().Be("a b c d e"); - concatString.Should().BeOfType(); - concatString.GetToolCalls().Should().HaveCount(1); - concatString.GetToolCalls().First().FunctionName.Should().Be(nameof(ConcatString)); - - var calculateTax = await agent.SendAsync("calculate tax: 100, 0.1"); - calculateTax.GetContent().Should().Be("tax is 10"); - calculateTax.Should().BeOfType(); - calculateTax.GetToolCalls().Should().HaveCount(1); - calculateTax.GetToolCalls().First().FunctionName.Should().Be(nameof(CalculateTax)); - - // parallel function calls - var calculateTaxes = await agent.SendAsync("calculate tax: 100, 0.1; calculate tax: 200, 0.2"); - calculateTaxes.GetContent().Should().Be("tax is 10\ntax is 40"); // "tax is 10\n tax is 40 - calculateTaxes.Should().BeOfType(); - calculateTaxes.GetToolCalls().Should().HaveCount(2); - calculateTaxes.GetToolCalls().First().FunctionName.Should().Be(nameof(CalculateTax)); - - // send aggregate message back to llm to get the final result - var finalResult = await agent.SendAsync(calculateTaxes); - } } diff --git a/dotnet/samples/AutoGen.BasicSamples/Program.cs b/dotnet/samples/AutoGen.BasicSamples/Program.cs index 16a79e75cffe..3a2edbb585f4 100644 --- a/dotnet/samples/AutoGen.BasicSamples/Program.cs +++ b/dotnet/samples/AutoGen.BasicSamples/Program.cs @@ -11,8 +11,7 @@ // When a new sample is created please add them to the allSamples collection ("Assistant Agent", Example01_AssistantAgent.RunAsync), ("Two-agent Math Chat", Example02_TwoAgent_MathChat.RunAsync), - ("Agent Function Call With Source Generator", Example03_Agent_FunctionCall.ToolCallWithSourceGenerator), - ("Agent Function Call With M.E.A.I AI Functions", Example03_Agent_FunctionCall.ToolCallWithMEAITools), + ("Agent Function Call", Example03_Agent_FunctionCall.RunAsync), ("Dynamic Group Chat Coding Task", Example04_Dynamic_GroupChat_Coding_Task.RunAsync), ("DALL-E and GPT4v", Example05_Dalle_And_GPT4V.RunAsync), ("User Proxy Agent", Example06_UserProxyAgent.RunAsync), diff --git a/dotnet/samples/AutoGen.WebAPI.Sample/Properties/launchSettings.json b/dotnet/samples/AutoGen.WebAPI.Sample/Properties/launchSettings.json deleted file mode 100644 index b9cc7582305f..000000000000 --- a/dotnet/samples/AutoGen.WebAPI.Sample/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "AutoGen.WebAPI.Sample": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:50675;http://localhost:50676" - } - } -} \ No newline at end of file diff --git a/dotnet/samples/Hello/Backend/Backend.csproj b/dotnet/samples/Hello/Backend/Backend.csproj index 2f5a02ee5117..60097b5d379d 100644 --- a/dotnet/samples/Hello/Backend/Backend.csproj +++ b/dotnet/samples/Hello/Backend/Backend.csproj @@ -1,16 +1,14 @@ - - - Exe - net8.0 - enable - enable - - + - + + Exe + net8.0 + enable + enable + diff --git a/dotnet/samples/Hello/Backend/Program.cs b/dotnet/samples/Hello/Backend/Program.cs index 7abdb205a85c..9f55daf69fc9 100644 --- a/dotnet/samples/Hello/Backend/Program.cs +++ b/dotnet/samples/Hello/Backend/Program.cs @@ -1,5 +1,7 @@ // Copyright (c) Microsoft Corporation. All rights reserved. // Program.cs +using Microsoft.Extensions.Hosting; + var app = await Microsoft.AutoGen.Runtime.Host.StartAsync(local: true); await app.WaitForShutdownAsync(); diff --git a/dotnet/samples/Hello/Backend/Properties/launchSettings.json b/dotnet/samples/Hello/Backend/Properties/launchSettings.json deleted file mode 100644 index db9c6bf2c316..000000000000 --- a/dotnet/samples/Hello/Backend/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "Backend": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:53071;http://localhost:53072" - } - } -} \ No newline at end of file diff --git a/dotnet/samples/Hello/Hello.AppHost/Hello.AppHost.csproj b/dotnet/samples/Hello/Hello.AppHost/Hello.AppHost.csproj index 88d23268c44d..3ecd30dee13a 100644 --- a/dotnet/samples/Hello/Hello.AppHost/Hello.AppHost.csproj +++ b/dotnet/samples/Hello/Hello.AppHost/Hello.AppHost.csproj @@ -1,4 +1,5 @@ + Exe net8.0 @@ -9,12 +10,12 @@ - - + + - + diff --git a/dotnet/samples/Hello/Hello.AppHost/Properties/launchSettings.json b/dotnet/samples/Hello/Hello.AppHost/Properties/launchSettings.json deleted file mode 100644 index ea78f2933fdb..000000000000 --- a/dotnet/samples/Hello/Hello.AppHost/Properties/launchSettings.json +++ /dev/null @@ -1,43 +0,0 @@ -{ - "profiles": { - "https": { - "commandName": "Project", - "launchBrowser": true, - "dotnetRunMessages": true, - "applicationUrl": "https://localhost:15887;http://localhost:15888", - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development", - "DOTNET_ENVIRONMENT": "Development", - //"DOTNET_DASHBOARD_OTLP_ENDPOINT_URL": "https://localhost:16037", - "DOTNET_DASHBOARD_OTLP_HTTP_ENDPOINT_URL": "https://localhost:16038", - "DOTNET_RESOURCE_SERVICE_ENDPOINT_URL": "https://localhost:17037", - "DOTNET_ASPIRE_SHOW_DASHBOARD_RESOURCES": "true" - } - }, - "http": { - "commandName": "Project", - "launchBrowser": true, - "dotnetRunMessages": true, - "applicationUrl": "http://localhost:15888", - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development", - "DOTNET_ENVIRONMENT": "Development", - //"DOTNET_DASHBOARD_OTLP_ENDPOINT_URL": "http://localhost:16031", - "DOTNET_DASHBOARD_OTLP_HTTP_ENDPOINT_URL": "http://localhost:16032", - "DOTNET_RESOURCE_SERVICE_ENDPOINT_URL": "http://localhost:17031", - "DOTNET_ASPIRE_SHOW_DASHBOARD_RESOURCES": "true", - "ASPIRE_ALLOW_UNSECURED_TRANSPORT": "true" - } - }, - "generate-manifest": { - "commandName": "Project", - "dotnetRunMessages": true, - "commandLineArgs": "--publisher manifest --output-path aspire-manifest.json", - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development", - "DOTNET_ENVIRONMENT": "Development" - } - } - }, - "$schema": "https://json.schemastore.org/launchsettings.json" -} diff --git a/dotnet/samples/Hello/HelloAIAgents/HelloAIAgent.cs b/dotnet/samples/Hello/HelloAIAgents/HelloAIAgent.cs index f7939da7d68e..ebde6d6d2f51 100644 --- a/dotnet/samples/Hello/HelloAIAgents/HelloAIAgent.cs +++ b/dotnet/samples/Hello/HelloAIAgents/HelloAIAgent.cs @@ -4,6 +4,7 @@ using Microsoft.AutoGen.Abstractions; using Microsoft.AutoGen.Agents; using Microsoft.Extensions.AI; +using Microsoft.Extensions.DependencyInjection; namespace Hello; [TopicSubscription("HelloAgents")] diff --git a/dotnet/samples/Hello/HelloAIAgents/HelloAIAgents.csproj b/dotnet/samples/Hello/HelloAIAgents/HelloAIAgents.csproj index 86bccb13b371..73f1891b3f22 100644 --- a/dotnet/samples/Hello/HelloAIAgents/HelloAIAgents.csproj +++ b/dotnet/samples/Hello/HelloAIAgents/HelloAIAgents.csproj @@ -1,14 +1,4 @@ - - - Exe - net8.0 - enable - enable - - - - - + @@ -16,4 +6,16 @@ + + + + + + + Exe + net8.0 + enable + enable + + diff --git a/dotnet/samples/Hello/HelloAIAgents/Program.cs b/dotnet/samples/Hello/HelloAIAgents/Program.cs index ebede82bb4fb..9d1964bfd1e1 100644 --- a/dotnet/samples/Hello/HelloAIAgents/Program.cs +++ b/dotnet/samples/Hello/HelloAIAgents/Program.cs @@ -2,8 +2,11 @@ // Program.cs using Hello; +using Microsoft.AspNetCore.Builder; using Microsoft.AutoGen.Abstractions; using Microsoft.AutoGen.Agents; +using Microsoft.Extensions.DependencyInjection; +using Microsoft.Extensions.Hosting; // send a message to the agent var builder = WebApplication.CreateBuilder(); diff --git a/dotnet/samples/Hello/HelloAIAgents/Properties/launchSettings.json b/dotnet/samples/Hello/HelloAIAgents/Properties/launchSettings.json deleted file mode 100644 index a5d241b0b325..000000000000 --- a/dotnet/samples/Hello/HelloAIAgents/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "HelloAIAgents": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:53139;http://localhost:53140" - } - } -} \ No newline at end of file diff --git a/dotnet/samples/Hello/HelloAgent/HelloAgent.csproj b/dotnet/samples/Hello/HelloAgent/HelloAgent.csproj index 8799eb7275d1..eb2ba96d6644 100644 --- a/dotnet/samples/Hello/HelloAgent/HelloAgent.csproj +++ b/dotnet/samples/Hello/HelloAgent/HelloAgent.csproj @@ -1,4 +1,16 @@ - + + + + + + + + + + + + + Exe net8.0 @@ -6,14 +18,4 @@ enable - - - - - - - - - - diff --git a/dotnet/samples/Hello/HelloAgent/Program.cs b/dotnet/samples/Hello/HelloAgent/Program.cs index 02ad838dea0d..fbe5d2f6dff9 100644 --- a/dotnet/samples/Hello/HelloAgent/Program.cs +++ b/dotnet/samples/Hello/HelloAgent/Program.cs @@ -3,6 +3,8 @@ using Microsoft.AutoGen.Abstractions; using Microsoft.AutoGen.Agents; +using Microsoft.Extensions.DependencyInjection; +using Microsoft.Extensions.Hosting; // step 1: create in-memory agent runtime @@ -25,8 +27,7 @@ namespace Hello [TopicSubscription("HelloAgents")] public class HelloAgent( IAgentContext context, - [FromKeyedServices("EventTypes")] EventTypes typeRegistry, - IHostApplicationLifetime hostApplicationLifetime) : AgentBase( + [FromKeyedServices("EventTypes")] EventTypes typeRegistry) : AgentBase( context, typeRegistry), ISayHello, @@ -57,11 +58,11 @@ public async Task Handle(ConversationClosed item) Message = goodbye }.ToCloudEvent(this.AgentId.Key); await PublishEvent(evt).ConfigureAwait(false); + //sleep + await Task.Delay(10000).ConfigureAwait(false); + await AgentsApp.ShutdownAsync().ConfigureAwait(false); - // Signal shutdown. - hostApplicationLifetime.StopApplication(); } - public async Task SayHello(string ask) { var response = $"\n\n\n\n***************Hello {ask}**********************\n\n\n\n"; diff --git a/dotnet/samples/Hello/HelloAgent/Properties/launchSettings.json b/dotnet/samples/Hello/HelloAgent/Properties/launchSettings.json deleted file mode 100644 index 04cd1b228704..000000000000 --- a/dotnet/samples/Hello/HelloAgent/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "HelloAgent": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:53113;http://localhost:53114" - } - } -} \ No newline at end of file diff --git a/dotnet/samples/Hello/HelloAgentState/HelloAgentState.csproj b/dotnet/samples/Hello/HelloAgentState/HelloAgentState.csproj index 8799eb7275d1..eb2ba96d6644 100644 --- a/dotnet/samples/Hello/HelloAgentState/HelloAgentState.csproj +++ b/dotnet/samples/Hello/HelloAgentState/HelloAgentState.csproj @@ -1,4 +1,16 @@ - + + + + + + + + + + + + + Exe net8.0 @@ -6,14 +18,4 @@ enable - - - - - - - - - - diff --git a/dotnet/samples/Hello/HelloAgentState/Program.cs b/dotnet/samples/Hello/HelloAgentState/Program.cs index c1e00e4d6322..66b888d6c46e 100644 --- a/dotnet/samples/Hello/HelloAgentState/Program.cs +++ b/dotnet/samples/Hello/HelloAgentState/Program.cs @@ -3,6 +3,8 @@ using Microsoft.AutoGen.Abstractions; using Microsoft.AutoGen.Agents; +using Microsoft.Extensions.DependencyInjection; +using Microsoft.Extensions.Hosting; // send a message to the agent var app = await AgentsApp.PublishMessageAsync("HelloAgents", new NewMessageReceived diff --git a/dotnet/samples/Hello/HelloAgentState/Properties/launchSettings.json b/dotnet/samples/Hello/HelloAgentState/Properties/launchSettings.json deleted file mode 100644 index 067d2fb83551..000000000000 --- a/dotnet/samples/Hello/HelloAgentState/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "HelloAgentState": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:53136;http://localhost:53137" - } - } -} \ No newline at end of file diff --git a/dotnet/samples/dev-team/DevTeam.AgentHost/Properties/launchSettings.json b/dotnet/samples/dev-team/DevTeam.AgentHost/Properties/launchSettings.json deleted file mode 100644 index c43e7586ac17..000000000000 --- a/dotnet/samples/dev-team/DevTeam.AgentHost/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "DevTeam.AgentHost": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:50670;http://localhost:50673" - } - } -} \ No newline at end of file diff --git a/dotnet/samples/dev-team/DevTeam.Agents/DevTeam.Agents.csproj b/dotnet/samples/dev-team/DevTeam.Agents/DevTeam.Agents.csproj index 03176d2cfd25..d7dbd1688599 100644 --- a/dotnet/samples/dev-team/DevTeam.Agents/DevTeam.Agents.csproj +++ b/dotnet/samples/dev-team/DevTeam.Agents/DevTeam.Agents.csproj @@ -7,9 +7,9 @@ - + - + diff --git a/dotnet/samples/dev-team/DevTeam.Agents/Properties/launchSettings.json b/dotnet/samples/dev-team/DevTeam.Agents/Properties/launchSettings.json deleted file mode 100644 index 8edfece6ad8d..000000000000 --- a/dotnet/samples/dev-team/DevTeam.Agents/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "DevTeam.Agents": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:50669;http://localhost:50671" - } - } -} \ No newline at end of file diff --git a/dotnet/samples/dev-team/DevTeam.Backend/Properties/launchSettings.json b/dotnet/samples/dev-team/DevTeam.Backend/Properties/launchSettings.json deleted file mode 100644 index f63e521d5545..000000000000 --- a/dotnet/samples/dev-team/DevTeam.Backend/Properties/launchSettings.json +++ /dev/null @@ -1,12 +0,0 @@ -{ - "profiles": { - "DevTeam.Backend": { - "commandName": "Project", - "launchBrowser": true, - "environmentVariables": { - "ASPNETCORE_ENVIRONMENT": "Development" - }, - "applicationUrl": "https://localhost:50672;http://localhost:50674" - } - } -} \ No newline at end of file diff --git a/dotnet/src/AutoGen.Core/AutoGen.Core.csproj b/dotnet/src/AutoGen.Core/AutoGen.Core.csproj index f46d48dc844f..c34c03af2b51 100644 --- a/dotnet/src/AutoGen.Core/AutoGen.Core.csproj +++ b/dotnet/src/AutoGen.Core/AutoGen.Core.csproj @@ -17,7 +17,6 @@ - diff --git a/dotnet/src/AutoGen.Core/Function/FunctionAttribute.cs b/dotnet/src/AutoGen.Core/Function/FunctionAttribute.cs index 9418dc7fd6ae..bb37f1cb25de 100644 --- a/dotnet/src/AutoGen.Core/Function/FunctionAttribute.cs +++ b/dotnet/src/AutoGen.Core/Function/FunctionAttribute.cs @@ -3,9 +3,6 @@ using System; using System.Collections.Generic; -using System.Linq; -using System.Text.Json.Serialization; -using Microsoft.Extensions.AI; namespace AutoGen.Core; @@ -25,10 +22,6 @@ public FunctionAttribute(string? functionName = null, string? description = null public class FunctionContract { - private const string NamespaceKey = nameof(Namespace); - - private const string ClassNameKey = nameof(ClassName); - /// /// The namespace of the function. /// @@ -59,7 +52,6 @@ public class FunctionContract /// /// The return type of the function. /// - [JsonIgnore] public Type? ReturnType { get; set; } /// @@ -68,39 +60,6 @@ public class FunctionContract /// Otherwise, the description will be null. /// public string? ReturnDescription { get; set; } - - public static implicit operator FunctionContract(AIFunctionMetadata metadata) - { - return new FunctionContract - { - Namespace = metadata.AdditionalProperties.ContainsKey(NamespaceKey) ? metadata.AdditionalProperties[NamespaceKey] as string : null, - ClassName = metadata.AdditionalProperties.ContainsKey(ClassNameKey) ? metadata.AdditionalProperties[ClassNameKey] as string : null, - Name = metadata.Name, - Description = metadata.Description, - Parameters = metadata.Parameters?.Select(p => (FunctionParameterContract)p).ToList(), - ReturnType = metadata.ReturnParameter.ParameterType, - ReturnDescription = metadata.ReturnParameter.Description, - }; - } - - public static implicit operator AIFunctionMetadata(FunctionContract contract) - { - return new AIFunctionMetadata(contract.Name) - { - Description = contract.Description, - ReturnParameter = new AIFunctionReturnParameterMetadata() - { - Description = contract.ReturnDescription, - ParameterType = contract.ReturnType, - }, - AdditionalProperties = new Dictionary - { - [NamespaceKey] = contract.Namespace, - [ClassNameKey] = contract.ClassName, - }, - Parameters = [.. contract.Parameters?.Select(p => (AIFunctionParameterMetadata)p)], - }; - } } public class FunctionParameterContract @@ -120,7 +79,6 @@ public class FunctionParameterContract /// /// The type of the parameter. /// - [JsonIgnore] public Type? ParameterType { get; set; } /// @@ -132,29 +90,4 @@ public class FunctionParameterContract /// The default value of the parameter. /// public object? DefaultValue { get; set; } - - // convert to/from FunctionParameterMetadata - public static implicit operator FunctionParameterContract(AIFunctionParameterMetadata metadata) - { - return new FunctionParameterContract - { - Name = metadata.Name, - Description = metadata.Description, - ParameterType = metadata.ParameterType, - IsRequired = metadata.IsRequired, - DefaultValue = metadata.DefaultValue, - }; - } - - public static implicit operator AIFunctionParameterMetadata(FunctionParameterContract contract) - { - return new AIFunctionParameterMetadata(contract.Name!) - { - DefaultValue = contract.DefaultValue, - Description = contract.Description, - IsRequired = contract.IsRequired, - ParameterType = contract.ParameterType, - HasDefaultValue = contract.DefaultValue != null, - }; - } } diff --git a/dotnet/src/AutoGen.Core/Middleware/FunctionCallMiddleware.cs b/dotnet/src/AutoGen.Core/Middleware/FunctionCallMiddleware.cs index 266155316c81..21461834dc83 100644 --- a/dotnet/src/AutoGen.Core/Middleware/FunctionCallMiddleware.cs +++ b/dotnet/src/AutoGen.Core/Middleware/FunctionCallMiddleware.cs @@ -5,10 +5,8 @@ using System.Collections.Generic; using System.Linq; using System.Runtime.CompilerServices; -using System.Text.Json; using System.Threading; using System.Threading.Tasks; -using Microsoft.Extensions.AI; namespace AutoGen.Core; @@ -45,19 +43,6 @@ public FunctionCallMiddleware( this.functionMap = functionMap; } - /// - /// Create a new instance of with a list of . - /// - /// function list - /// optional middleware name. If not provided, the class name will be used. - public FunctionCallMiddleware(IEnumerable functions, string? name = null) - { - this.Name = name ?? nameof(FunctionCallMiddleware); - this.functions = functions.Select(f => (FunctionContract)f.Metadata).ToArray(); - - this.functionMap = functions.Select(f => (f.Metadata.Name, this.AIToolInvokeWrapper(f.InvokeAsync))).ToDictionary(f => f.Name, f => f.Item2); - } - public string? Name { get; } public async Task InvokeAsync(MiddlewareContext context, IAgent agent, CancellationToken cancellationToken = default) @@ -188,20 +173,4 @@ private async Task InvokeToolCallMessagesAfterInvokingAgentAsync(ToolC return toolCallMsg; } } - - private Func> AIToolInvokeWrapper(Func>?, CancellationToken, Task> lambda) - { - return async (string args) => - { - var arguments = JsonSerializer.Deserialize>(args); - var result = await lambda(arguments, CancellationToken.None); - - return result switch - { - string s => s, - JsonElement e => e.ToString(), - _ => JsonSerializer.Serialize(result), - }; - }; - } } diff --git a/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentContext.cs b/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentContext.cs index ab5972730fb7..d93b6246765d 100644 --- a/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentContext.cs +++ b/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentContext.cs @@ -12,9 +12,9 @@ public interface IAgentContext IAgentBase? AgentInstance { get; set; } DistributedContextPropagator DistributedContextPropagator { get; } // TODO: Remove this. An abstraction should not have a dependency on DistributedContextPropagator. ILogger Logger { get; } // TODO: Remove this. An abstraction should not have a dependency on ILogger. - ValueTask Store(AgentState value, CancellationToken cancellationToken = default); - ValueTask Read(AgentId agentId, CancellationToken cancellationToken = default); - ValueTask SendResponseAsync(RpcRequest request, RpcResponse response, CancellationToken cancellationToken = default); - ValueTask SendRequestAsync(IAgentBase agent, RpcRequest request, CancellationToken cancellationToken = default); - ValueTask PublishEventAsync(CloudEvent @event, CancellationToken cancellationToken = default); + ValueTask Store(AgentState value); + ValueTask Read(AgentId agentId); + ValueTask SendResponseAsync(RpcRequest request, RpcResponse response); + ValueTask SendRequestAsync(IAgentBase agent, RpcRequest request); + ValueTask PublishEventAsync(CloudEvent @event); } diff --git a/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentWorkerRuntime.cs b/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentWorkerRuntime.cs index c03259f722f3..1a255e132346 100644 --- a/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentWorkerRuntime.cs +++ b/dotnet/src/Microsoft.AutoGen/Abstractions/IAgentWorkerRuntime.cs @@ -5,9 +5,9 @@ namespace Microsoft.AutoGen.Abstractions; public interface IAgentWorkerRuntime { - ValueTask PublishEvent(CloudEvent evt, CancellationToken cancellationToken); - ValueTask SendRequest(IAgentBase agent, RpcRequest request, CancellationToken cancellationToken); - ValueTask SendResponse(RpcResponse response, CancellationToken cancellationToken); - ValueTask Store(AgentState value, CancellationToken cancellationToken); - ValueTask Read(AgentId agentId, CancellationToken cancellationToken); + ValueTask PublishEvent(CloudEvent evt); + ValueTask SendRequest(IAgentBase agent, RpcRequest request); + ValueTask SendResponse(RpcResponse response); + ValueTask Store(AgentState value); + ValueTask Read(AgentId agentId); } diff --git a/dotnet/src/Microsoft.AutoGen/Agents/AgentBase.cs b/dotnet/src/Microsoft.AutoGen/Agents/AgentBase.cs index baa7ee201edd..af06c84e9ba1 100644 --- a/dotnet/src/Microsoft.AutoGen/Agents/AgentBase.cs +++ b/dotnet/src/Microsoft.AutoGen/Agents/AgentBase.cs @@ -202,14 +202,15 @@ public async ValueTask PublishEvent(CloudEvent item) var activity = s_source.StartActivity($"PublishEvent '{item.Type}'", ActivityKind.Client, Activity.Current?.Context ?? default); activity?.SetTag("peer.service", $"{item.Type}/{item.Source}"); + var completion = new TaskCompletionSource(TaskCreationOptions.RunContinuationsAsynchronously); // TODO: fix activity Context.DistributedContextPropagator.Inject(activity, item.Metadata, static (carrier, key, value) => ((IDictionary)carrier!)[key] = value); await this.InvokeWithActivityAsync( - static async ((AgentBase Agent, CloudEvent Event) state) => + static async ((AgentBase Agent, CloudEvent Event, TaskCompletionSource) state) => { await state.Agent._context.PublishEventAsync(state.Event).ConfigureAwait(false); }, - (this, item), + (this, item, completion), activity, item.Type).ConfigureAwait(false); } diff --git a/dotnet/src/Microsoft.AutoGen/Agents/AgentContext.cs b/dotnet/src/Microsoft.AutoGen/Agents/AgentContext.cs index 7de1e6565d33..325bc33a11d0 100644 --- a/dotnet/src/Microsoft.AutoGen/Agents/AgentContext.cs +++ b/dotnet/src/Microsoft.AutoGen/Agents/AgentContext.cs @@ -15,25 +15,25 @@ internal sealed class AgentContext(AgentId agentId, IAgentWorkerRuntime runtime, public ILogger Logger { get; } = logger; public IAgentBase? AgentInstance { get; set; } public DistributedContextPropagator DistributedContextPropagator { get; } = distributedContextPropagator; - public async ValueTask SendResponseAsync(RpcRequest request, RpcResponse response, CancellationToken cancellationToken) + public async ValueTask SendResponseAsync(RpcRequest request, RpcResponse response) { response.RequestId = request.RequestId; - await _runtime.SendResponse(response, cancellationToken).ConfigureAwait(false); + await _runtime.SendResponse(response); } - public async ValueTask SendRequestAsync(IAgentBase agent, RpcRequest request, CancellationToken cancellationToken) + public async ValueTask SendRequestAsync(IAgentBase agent, RpcRequest request) { - await _runtime.SendRequest(agent, request, cancellationToken).ConfigureAwait(false); + await _runtime.SendRequest(agent, request).ConfigureAwait(false); } - public async ValueTask PublishEventAsync(CloudEvent @event, CancellationToken cancellationToken) + public async ValueTask PublishEventAsync(CloudEvent @event) { - await _runtime.PublishEvent(@event, cancellationToken).ConfigureAwait(false); + await _runtime.PublishEvent(@event).ConfigureAwait(false); } - public async ValueTask Store(AgentState value, CancellationToken cancellationToken) + public async ValueTask Store(AgentState value) { - await _runtime.Store(value, cancellationToken).ConfigureAwait(false); + await _runtime.Store(value).ConfigureAwait(false); } - public ValueTask Read(AgentId agentId, CancellationToken cancellationToken) + public ValueTask Read(AgentId agentId) { - return _runtime.Read(agentId, cancellationToken); + return _runtime.Read(agentId); } } diff --git a/dotnet/src/Microsoft.AutoGen/Agents/GrpcAgentWorkerRuntime.cs b/dotnet/src/Microsoft.AutoGen/Agents/GrpcAgentWorkerRuntime.cs index b0550c1fb715..c52509876ffd 100644 --- a/dotnet/src/Microsoft.AutoGen/Agents/GrpcAgentWorkerRuntime.cs +++ b/dotnet/src/Microsoft.AutoGen/Agents/GrpcAgentWorkerRuntime.cs @@ -19,7 +19,7 @@ public sealed class GrpcAgentWorkerRuntime : IHostedService, IDisposable, IAgent private readonly ConcurrentDictionary _agentTypes = new(); private readonly ConcurrentDictionary<(string Type, string Key), IAgentBase> _agents = new(); private readonly ConcurrentDictionary _pendingRequests = new(); - private readonly Channel<(Message Message, TaskCompletionSource WriteCompletionSource)> _outboundMessagesChannel = Channel.CreateBounded<(Message, TaskCompletionSource)>(new BoundedChannelOptions(1024) + private readonly Channel _outboundMessagesChannel = Channel.CreateBounded(new BoundedChannelOptions(1024) { AllowSynchronousContinuations = true, SingleReader = true, @@ -138,34 +138,30 @@ private async Task RunWritePump() var outboundMessages = _outboundMessagesChannel.Reader; while (!_shutdownCts.IsCancellationRequested) { - (Message Message, TaskCompletionSource WriteCompletionSource) item = default; try { await outboundMessages.WaitToReadAsync().ConfigureAwait(false); // Read the next message if we don't already have an unsent message // waiting to be sent. - if (!outboundMessages.TryRead(out item)) + if (!outboundMessages.TryRead(out var message)) { break; } while (!_shutdownCts.IsCancellationRequested) { - await channel.RequestStream.WriteAsync(item.Message, _shutdownCts.Token).ConfigureAwait(false); - item.WriteCompletionSource.TrySetResult(); + await channel.RequestStream.WriteAsync(message, _shutdownCts.Token).ConfigureAwait(false); break; } } catch (OperationCanceledException) { // Time to shut down. - item.WriteCompletionSource?.TrySetCanceled(); break; } catch (Exception ex) when (!_shutdownCts.IsCancellationRequested) { - item.WriteCompletionSource?.TrySetException(ex); _logger.LogError(ex, "Error writing to channel."); channel = RecreateChannel(channel); continue; @@ -173,15 +169,9 @@ private async Task RunWritePump() catch { // Shutdown requested. - item.WriteCompletionSource?.TrySetCanceled(); break; } } - - while (outboundMessages.TryRead(out var item)) - { - item.WriteCompletionSource.TrySetCanceled(); - } } private IAgentBase GetOrActivateAgent(AgentId agentId) @@ -223,53 +213,33 @@ await WriteChannelAsync(new Message //StateType = state?.Name, //Events = { events } } - }, - _shutdownCts.Token).ConfigureAwait(false); + }).ConfigureAwait(false); } } - public async ValueTask SendResponse(RpcResponse response, CancellationToken cancellationToken) + public async ValueTask SendResponse(RpcResponse response) { _logger.LogInformation("Sending response '{Response}'.", response); - await WriteChannelAsync(new Message { Response = response }, cancellationToken).ConfigureAwait(false); + await WriteChannelAsync(new Message { Response = response }).ConfigureAwait(false); } - public async ValueTask SendRequest(IAgentBase agent, RpcRequest request, CancellationToken cancellationToken) + public async ValueTask SendRequest(IAgentBase agent, RpcRequest request) { _logger.LogInformation("[{AgentId}] Sending request '{Request}'.", agent.AgentId, request); var requestId = Guid.NewGuid().ToString(); _pendingRequests[requestId] = (agent, request.RequestId); request.RequestId = requestId; - try - { - await WriteChannelAsync(new Message { Request = request }, cancellationToken).ConfigureAwait(false); - } - catch (Exception exception) - { - if (_pendingRequests.TryRemove(requestId, out _)) - { - agent.ReceiveMessage(new Message { Response = new RpcResponse { RequestId = request.RequestId, Error = exception.Message } }); - } - } + await WriteChannelAsync(new Message { Request = request }).ConfigureAwait(false); } - public async ValueTask PublishEvent(CloudEvent @event, CancellationToken cancellationToken) + public async ValueTask PublishEvent(CloudEvent @event) { - try - { - await WriteChannelAsync(new Message { CloudEvent = @event }, cancellationToken).ConfigureAwait(false); - } - catch (Exception exception) - { - _logger.LogWarning(exception, "Failed to publish event '{Event}'.", @event); - } + await WriteChannelAsync(new Message { CloudEvent = @event }).ConfigureAwait(false); } - private async Task WriteChannelAsync(Message message, CancellationToken cancellationToken) + private async Task WriteChannelAsync(Message message) { - var tcs = new TaskCompletionSource(TaskCreationOptions.RunContinuationsAsynchronously); - await _outboundMessagesChannel.Writer.WriteAsync((message, tcs), cancellationToken).ConfigureAwait(false); - await tcs.Task.WaitAsync(cancellationToken); + await _outboundMessagesChannel.Writer.WriteAsync(message).ConfigureAwait(false); } private AsyncDuplexStreamingCall GetChannel() @@ -299,7 +269,7 @@ private AsyncDuplexStreamingCall RecreateChannel(AsyncDuplexSt if (_channel is null || _channel == channel) { _channel?.Dispose(); - _channel = _client.OpenChannel(cancellationToken: _shutdownCts.Token); + _channel = _client.OpenChannel(); } } } @@ -364,19 +334,19 @@ public async Task StopAsync(CancellationToken cancellationToken) _channel?.Dispose(); } } - public ValueTask Store(AgentState value, CancellationToken cancellationToken) + public ValueTask Store(AgentState value) { var agentId = value.AgentId ?? throw new InvalidOperationException("AgentId is required when saving AgentState."); - var response = _client.SaveState(value, cancellationToken: cancellationToken); + var response = _client.SaveState(value); if (!response.Success) { throw new InvalidOperationException($"Error saving AgentState for AgentId {agentId}."); } return ValueTask.CompletedTask; } - public async ValueTask Read(AgentId agentId, CancellationToken cancellationToken) + public async ValueTask Read(AgentId agentId) { - var response = await _client.GetStateAsync(agentId, cancellationToken: cancellationToken); + var response = await _client.GetStateAsync(agentId); // if (response.Success && response.AgentState.AgentId is not null) - why is success always false? if (response.AgentState.AgentId is not null) { diff --git a/dotnet/src/Microsoft.AutoGen/Extensions/AIModelClientHostingExtensions/AIModelClientHostingExtensions.csproj b/dotnet/src/Microsoft.AutoGen/Extensions/AIModelClientHostingExtensions/AIModelClientHostingExtensions.csproj index 2358351deb6c..a94921946c09 100644 --- a/dotnet/src/Microsoft.AutoGen/Extensions/AIModelClientHostingExtensions/AIModelClientHostingExtensions.csproj +++ b/dotnet/src/Microsoft.AutoGen/Extensions/AIModelClientHostingExtensions/AIModelClientHostingExtensions.csproj @@ -14,6 +14,5 @@ - diff --git a/dotnet/src/Microsoft.AutoGen/Extensions/SemanticKernel/Microsoft.AutoGen.Extensions.SemanticKernel.csproj b/dotnet/src/Microsoft.AutoGen/Extensions/SemanticKernel/Microsoft.AutoGen.Extensions.SemanticKernel.csproj index fb47750fd44d..a976c007715c 100644 --- a/dotnet/src/Microsoft.AutoGen/Extensions/SemanticKernel/Microsoft.AutoGen.Extensions.SemanticKernel.csproj +++ b/dotnet/src/Microsoft.AutoGen/Extensions/SemanticKernel/Microsoft.AutoGen.Extensions.SemanticKernel.csproj @@ -14,7 +14,7 @@ - + diff --git a/dotnet/test/AutoGen.Tests/AutoGen.Tests.csproj b/dotnet/test/AutoGen.Tests/AutoGen.Tests.csproj index 248a9e29b00d..12c31e1a473c 100644 --- a/dotnet/test/AutoGen.Tests/AutoGen.Tests.csproj +++ b/dotnet/test/AutoGen.Tests/AutoGen.Tests.csproj @@ -13,8 +13,6 @@ - - diff --git a/dotnet/test/AutoGen.Tests/BasicSampleTest.cs b/dotnet/test/AutoGen.Tests/BasicSampleTest.cs index 5cf4704037fc..df02bb3dcd0f 100644 --- a/dotnet/test/AutoGen.Tests/BasicSampleTest.cs +++ b/dotnet/test/AutoGen.Tests/BasicSampleTest.cs @@ -34,8 +34,7 @@ public async Task TwoAgentMathClassTestAsync() [ApiKeyFact("OPENAI_API_KEY")] public async Task AgentFunctionCallTestAsync() { - await Example03_Agent_FunctionCall.ToolCallWithSourceGenerator(); - await Example03_Agent_FunctionCall.ToolCallWithMEAITools(); + await Example03_Agent_FunctionCall.RunAsync(); } [ApiKeyFact("MISTRAL_API_KEY")] diff --git a/dotnet/test/AutoGen.Tests/Function/ApprovalTests/FunctionTests.CreateGetWeatherFunctionFromAIFunctionFactoryAsync.approved.txt b/dotnet/test/AutoGen.Tests/Function/ApprovalTests/FunctionTests.CreateGetWeatherFunctionFromAIFunctionFactoryAsync.approved.txt deleted file mode 100644 index f57e0203e353..000000000000 --- a/dotnet/test/AutoGen.Tests/Function/ApprovalTests/FunctionTests.CreateGetWeatherFunctionFromAIFunctionFactoryAsync.approved.txt +++ /dev/null @@ -1,76 +0,0 @@ -[ - { - "Kind": 0, - "FunctionName": "GetWeather", - "FunctionDescription": "get weather", - "FunctionParameters": { - "type": "object", - "properties": { - "city": { - "type": "string" - }, - "date": { - "type": "string" - } - }, - "required": [ - "city" - ] - } - }, - { - "Kind": 0, - "FunctionName": "GetWeatherStatic", - "FunctionDescription": "get weather from static method", - "FunctionParameters": { - "type": "object", - "properties": { - "city": { - "type": "string" - }, - "date": { - "type": "array", - "items": { - "type": "string" - } - } - }, - "required": [ - "city", - "date" - ] - } - }, - { - "Kind": 0, - "FunctionName": "GetWeather", - "FunctionDescription": "get weather from async method", - "FunctionParameters": { - "type": "object", - "properties": { - "city": { - "type": "string" - } - }, - "required": [ - "city" - ] - } - }, - { - "Kind": 0, - "FunctionName": "GetWeatherAsyncStatic", - "FunctionDescription": "get weather from async static method", - "FunctionParameters": { - "type": "object", - "properties": { - "city": { - "type": "string" - } - }, - "required": [ - "city" - ] - } - } -] \ No newline at end of file diff --git a/dotnet/test/AutoGen.Tests/Function/FunctionTests.cs b/dotnet/test/AutoGen.Tests/Function/FunctionTests.cs deleted file mode 100644 index 64abb293bb16..000000000000 --- a/dotnet/test/AutoGen.Tests/Function/FunctionTests.cs +++ /dev/null @@ -1,82 +0,0 @@ -// Copyright (c) Microsoft Corporation. All rights reserved. -// FunctionTests.cs - -using System; -using System.ComponentModel; -using System.Linq; -using System.Text.Json; -using System.Text.Json.Serialization; -using System.Threading.Tasks; -using ApprovalTests; -using ApprovalTests.Namers; -using ApprovalTests.Reporters; -using AutoGen.OpenAI.Extension; -using FluentAssertions; -using Microsoft.Extensions.AI; -using Xunit; - -namespace AutoGen.Tests.Function; -public class FunctionTests -{ - private readonly JsonSerializerOptions _jsonSerializerOptions = new() { WriteIndented = true, DefaultIgnoreCondition = JsonIgnoreCondition.WhenWritingNull }; - [Description("get weather")] - public string GetWeather(string city, string date = "today") - { - return $"The weather in {city} is sunny."; - } - - [Description("get weather from static method")] - [return: Description("weather information")] - public static string GetWeatherStatic(string city, string[] date) - { - return $"The weather in {city} is sunny."; - } - - [Description("get weather from async method")] - public async Task GetWeatherAsync(string city) - { - await Task.Delay(100); - return $"The weather in {city} is sunny."; - } - - [Description("get weather from async static method")] - public static async Task GetWeatherAsyncStatic(string city) - { - await Task.Delay(100); - return $"The weather in {city} is sunny."; - } - - [Fact] - [UseReporter(typeof(DiffReporter))] - [UseApprovalSubdirectory("ApprovalTests")] - public async Task CreateGetWeatherFunctionFromAIFunctionFactoryAsync() - { - Delegate[] availableDelegates = [ - GetWeather, - GetWeatherStatic, - GetWeatherAsync, - GetWeatherAsyncStatic, - ]; - - var functionContracts = availableDelegates.Select(function => (FunctionContract)AIFunctionFactory.Create(function).Metadata).ToList(); - - // Verify the function contracts - functionContracts.Should().HaveCount(4); - - var openAIToolContracts = functionContracts.Select(f => - { - var tool = f.ToChatTool(); - - return new - { - tool.Kind, - tool.FunctionName, - tool.FunctionDescription, - FunctionParameters = tool.FunctionParameters.ToObjectFromJson(), - }; - }); - - var json = JsonSerializer.Serialize(openAIToolContracts, _jsonSerializerOptions); - Approvals.Verify(json); - } -} diff --git a/dotnet/test/AutoGen.Tests/MiddlewareTest.cs b/dotnet/test/AutoGen.Tests/MiddlewareTest.cs index 61691b225437..d98fa14ec19d 100644 --- a/dotnet/test/AutoGen.Tests/MiddlewareTest.cs +++ b/dotnet/test/AutoGen.Tests/MiddlewareTest.cs @@ -7,7 +7,6 @@ using System.Text.Json; using System.Threading.Tasks; using FluentAssertions; -using Microsoft.Extensions.AI; using Xunit; namespace AutoGen.Tests; @@ -73,7 +72,7 @@ public async Task FunctionCallMiddlewareTestAsync() var agent = new EchoAgent("echo"); var args = new EchoSchema { message = "hello" }; var argsJson = JsonSerializer.Serialize(args) ?? throw new InvalidOperationException("Failed to serialize args"); - var functionCall = new ToolCall("Echo", argsJson); + var functionCall = new ToolCall("echo", argsJson); var functionCallAgent = agent.RegisterMiddleware(async (messages, options, agent, ct) => { if (options?.Functions is null) @@ -87,7 +86,7 @@ public async Task FunctionCallMiddlewareTestAsync() // test 1 // middleware should invoke function call if the message is a function call message var mw = new FunctionCallMiddleware( - functionMap: new Dictionary>> { { "Echo", EchoWrapper } }); + functionMap: new Dictionary>> { { "echo", EchoWrapper } }); var testAgent = agent.RegisterMiddleware(mw); var functionCallMessage = new ToolCallMessage(functionCall.FunctionName, functionCall.FunctionArguments, from: "user"); @@ -97,38 +96,30 @@ public async Task FunctionCallMiddlewareTestAsync() reply.From.Should().Be("echo"); // test 2 - // middleware should work with AIFunction from M.E.A.I - var getWeatherTool = AIFunctionFactory.Create(this.Echo); - mw = new FunctionCallMiddleware([getWeatherTool]); - testAgent = agent.RegisterMiddleware(mw); - reply = await testAgent.SendAsync(functionCallMessage); - reply.GetContent()!.Should().Be("[FUNC] hello"); - - // test 3 // middleware should invoke function call if agent reply is a function call message mw = new FunctionCallMiddleware( functions: [this.EchoFunctionContract], - functionMap: new Dictionary>> { { "Echo", EchoWrapper } }); + functionMap: new Dictionary>> { { "echo", EchoWrapper } }); testAgent = functionCallAgent.RegisterMiddleware(mw); reply = await testAgent.SendAsync("hello"); reply.GetContent()!.Should().Be("[FUNC] hello"); reply.From.Should().Be("echo"); - // test 4 + // test 3 // middleware should return original reply if the reply from agent is not a function call message mw = new FunctionCallMiddleware( - functionMap: new Dictionary>> { { "Echo", EchoWrapper } }); + functionMap: new Dictionary>> { { "echo", EchoWrapper } }); testAgent = agent.RegisterMiddleware(mw); reply = await testAgent.SendAsync("hello"); reply.GetContent()!.Should().Be("hello"); reply.From.Should().Be("echo"); - // test 5 + // test 4 // middleware should return an error message if the function name is not available when invoking the function from previous agent reply mw = new FunctionCallMiddleware( - functionMap: new Dictionary>> { { "Echo2", EchoWrapper } }); + functionMap: new Dictionary>> { { "echo2", EchoWrapper } }); testAgent = agent.RegisterMiddleware(mw); reply = await testAgent.SendAsync(functionCallMessage); - reply.GetContent()!.Should().Be("Function Echo is not available. Available functions are: Echo2"); + reply.GetContent()!.Should().Be("Function echo is not available. Available functions are: echo2"); } } diff --git a/dotnet/website/articles/Create-type-safe-function-call.md b/dotnet/website/articles/Create-type-safe-function-call.md index 059a912d0717..a12869661817 100644 --- a/dotnet/website/articles/Create-type-safe-function-call.md +++ b/dotnet/website/articles/Create-type-safe-function-call.md @@ -1,4 +1,4 @@ -## Create type-safe function call using AutoGen.SourceGenerator +## Type-safe function call `AutoGen` provides a source generator to easness the trouble of manually craft function definition and function call wrapper from a function. To use this feature, simply add the `AutoGen.SourceGenerator` package to your project and decorate your function with @AutoGen.Core.FunctionAttribute. diff --git a/python/packages/autogen-agentchat/pyproject.toml b/python/packages/autogen-agentchat/pyproject.toml index 755959b5abf8..48874b9de662 100644 --- a/python/packages/autogen-agentchat/pyproject.toml +++ b/python/packages/autogen-agentchat/pyproject.toml @@ -4,7 +4,7 @@ build-backend = "hatchling.build" [project] name = "autogen-agentchat" -version = "0.4.0.dev3" +version = "0.4.0.dev2" license = {file = "LICENSE-CODE"} description = "AutoGen agents and teams library" readme = "README.md" @@ -15,7 +15,7 @@ classifiers = [ "Operating System :: OS Independent", ] dependencies = [ - "autogen-core==0.4.0.dev3", + "autogen-core==0.4.0.dev2", ] [tool.uv] diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_assistant_agent.py b/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_assistant_agent.py index daaacda15870..5414f782f022 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_assistant_agent.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_assistant_agent.py @@ -1,7 +1,7 @@ import asyncio import json import logging -from typing import Any, AsyncGenerator, Awaitable, Callable, Dict, List, Sequence +from typing import Any, Awaitable, Callable, Dict, List, Sequence from autogen_core.base import CancellationToken from autogen_core.components import FunctionCall @@ -24,9 +24,10 @@ HandoffMessage, InnerMessage, ResetMessage, + StopMessage, TextMessage, ToolCallMessage, - ToolCallResultMessage, + ToolCallResultMessages, ) from ._base_chat_agent import BaseChatAgent @@ -97,11 +98,7 @@ def set_defaults(cls, values: Dict[str, Any]) -> Dict[str, Any]: @property def handoff_tool(self) -> Tool: """Create a handoff tool from this handoff configuration.""" - - def _handoff_tool() -> str: - return self.message - - return FunctionTool(_handoff_tool, name=self.name, description=self.description) + return FunctionTool(lambda: self.message, name=self.name, description=self.description) class AssistantAgent(BaseChatAgent): @@ -130,27 +127,21 @@ class AssistantAgent(BaseChatAgent): .. code-block:: python - import asyncio from autogen_ext.models import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.task import MaxMessageTermination - async def main() -> None: - model_client = OpenAIChatCompletionClient(model="gpt-4o") - agent = AssistantAgent(name="assistant", model_client=model_client) + model_client = OpenAIChatCompletionClient(model="gpt-4o") + agent = AssistantAgent(name="assistant", model_client=model_client) - result await agent.run("What is the capital of France?", termination_condition=MaxMessageTermination(2)) - print(result) - - asyncio.run(main()) + await agent.run("What is the capital of France?", termination_condition=MaxMessageTermination(2)) The following example demonstrates how to create an assistant agent with - a model client and a tool, and generate a stream of messages for a task. + a model client and a tool, and generate a response to a simple task using the tool. .. code-block:: python - import asyncio from autogen_ext.models import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.task import MaxMessageTermination @@ -160,17 +151,10 @@ async def get_current_time() -> str: return "The current time is 12:00 PM." - async def main() -> None: - model_client = OpenAIChatCompletionClient(model="gpt-4o") - agent = AssistantAgent(name="assistant", model_client=model_client, tools=[get_current_time]) - - stream = agent.run_stream("What is the current time?", termination_condition=MaxMessageTermination(3)) + model_client = OpenAIChatCompletionClient(model="gpt-4o") + agent = AssistantAgent(name="assistant", model_client=model_client, tools=[get_current_time]) - async for message in stream: - print(message) - - - asyncio.run(main()) + await agent.run("What is the current time?", termination_condition=MaxMessageTermination(3)) """ @@ -231,18 +215,10 @@ def __init__( def produced_message_types(self) -> List[type[ChatMessage]]: """The types of messages that the assistant agent produces.""" if self._handoffs: - return [TextMessage, HandoffMessage] - return [TextMessage] + return [TextMessage, HandoffMessage, StopMessage] + return [TextMessage, StopMessage] async def on_messages(self, messages: Sequence[ChatMessage], cancellation_token: CancellationToken) -> Response: - async for message in self.on_messages_stream(messages, cancellation_token): - if isinstance(message, Response): - return message - raise AssertionError("The stream should have returned the final result.") - - async def on_messages_stream( - self, messages: Sequence[ChatMessage], cancellation_token: CancellationToken - ) -> AsyncGenerator[InnerMessage | Response, None]: # Add messages to the model context. for msg in messages: if isinstance(msg, ResetMessage): @@ -266,8 +242,7 @@ async def on_messages_stream( while isinstance(result.content, list) and all(isinstance(item, FunctionCall) for item in result.content): event_logger.debug(ToolCallEvent(tool_calls=result.content, source=self.name)) # Add the tool call message to the output. - inner_messages.append(ToolCallMessage(content=result.content, source=self.name, models_usage=result.usage)) - yield ToolCallMessage(content=result.content, source=self.name, models_usage=result.usage) + inner_messages.append(ToolCallMessage(content=result.content, source=self.name)) # Execute the tool calls. results = await asyncio.gather( @@ -275,8 +250,7 @@ async def on_messages_stream( ) event_logger.debug(ToolCallResultEvent(tool_call_results=results, source=self.name)) self._model_context.append(FunctionExecutionResultMessage(content=results)) - inner_messages.append(ToolCallResultMessage(content=results, source=self.name)) - yield ToolCallResultMessage(content=results, source=self.name) + inner_messages.append(ToolCallResultMessages(content=results, source=self.name)) # Detect handoff requests. handoffs: List[Handoff] = [] @@ -287,13 +261,12 @@ async def on_messages_stream( if len(handoffs) > 1: raise ValueError(f"Multiple handoffs detected: {[handoff.name for handoff in handoffs]}") # Return the output messages to signal the handoff. - yield Response( + return Response( chat_message=HandoffMessage( content=handoffs[0].message, target=handoffs[0].target, source=self.name ), inner_messages=inner_messages, ) - return # Generate an inference result based on the current model context. result = await self._model_client.create( @@ -302,9 +275,15 @@ async def on_messages_stream( self._model_context.append(AssistantMessage(content=result.content, source=self.name)) assert isinstance(result.content, str) - yield Response( - chat_message=TextMessage(content=result.content, source=self.name, models_usage=result.usage), - inner_messages=inner_messages, + # Detect stop request. + request_stop = "terminate" in result.content.strip().lower() + if request_stop: + return Response( + chat_message=StopMessage(content=result.content, source=self.name), inner_messages=inner_messages + ) + + return Response( + chat_message=TextMessage(content=result.content, source=self.name), inner_messages=inner_messages ) async def _execute_tool_call( diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_base_chat_agent.py b/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_base_chat_agent.py index cf146b0c10fb..ac74077e27c8 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_base_chat_agent.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/agents/_base_chat_agent.py @@ -1,9 +1,9 @@ from abc import ABC, abstractmethod -from typing import AsyncGenerator, List, Sequence +from typing import List, Sequence from autogen_core.base import CancellationToken -from ..base import ChatAgent, Response, TaskResult +from ..base import ChatAgent, Response, TaskResult, TerminationCondition from ..messages import ChatMessage, InnerMessage, TextMessage @@ -40,22 +40,12 @@ async def on_messages(self, messages: Sequence[ChatMessage], cancellation_token: """Handles incoming messages and returns a response.""" ... - async def on_messages_stream( - self, messages: Sequence[ChatMessage], cancellation_token: CancellationToken - ) -> AsyncGenerator[InnerMessage | Response, None]: - """Handles incoming messages and returns a stream of messages and - and the final item is the response. The base implementation in :class:`BaseChatAgent` - simply calls :meth:`on_messages` and yields the messages in the response.""" - response = await self.on_messages(messages, cancellation_token) - for inner_message in response.inner_messages or []: - yield inner_message - yield response - async def run( self, task: str, *, cancellation_token: CancellationToken | None = None, + termination_condition: TerminationCondition | None = None, ) -> TaskResult: """Run the agent with the given task and return the result.""" if cancellation_token is None: @@ -67,25 +57,3 @@ async def run( messages += response.inner_messages messages.append(response.chat_message) return TaskResult(messages=messages) - - async def run_stream( - self, - task: str, - *, - cancellation_token: CancellationToken | None = None, - ) -> AsyncGenerator[InnerMessage | ChatMessage | TaskResult, None]: - """Run the agent with the given task and return a stream of messages - and the final task result as the last item in the stream.""" - if cancellation_token is None: - cancellation_token = CancellationToken() - first_message = TextMessage(content=task, source="user") - yield first_message - messages: List[InnerMessage | ChatMessage] = [first_message] - async for message in self.on_messages_stream([first_message], cancellation_token): - if isinstance(message, Response): - yield message.chat_message - messages.append(message.chat_message) - yield TaskResult(messages=messages) - else: - messages.append(message) - yield message diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_chat_agent.py b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_chat_agent.py index ce73352daecc..d60dba349cbb 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_chat_agent.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_chat_agent.py @@ -1,10 +1,11 @@ from dataclasses import dataclass -from typing import AsyncGenerator, List, Protocol, Sequence, runtime_checkable +from typing import List, Protocol, Sequence, runtime_checkable from autogen_core.base import CancellationToken from ..messages import ChatMessage, InnerMessage -from ._task import TaskRunner +from ._task import TaskResult, TaskRunner +from ._termination import TerminationCondition @dataclass(kw_only=True) @@ -44,9 +45,12 @@ async def on_messages(self, messages: Sequence[ChatMessage], cancellation_token: """Handles incoming messages and returns a response.""" ... - def on_messages_stream( - self, messages: Sequence[ChatMessage], cancellation_token: CancellationToken - ) -> AsyncGenerator[InnerMessage | Response, None]: - """Handles incoming messages and returns a stream of inner messages and - and the final item is the response.""" + async def run( + self, + task: str, + *, + cancellation_token: CancellationToken | None = None, + termination_condition: TerminationCondition | None = None, + ) -> TaskResult: + """Run the agent with the given task and return the result.""" ... diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_task.py b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_task.py index a642120799f2..326cceecb1fd 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_task.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_task.py @@ -1,9 +1,10 @@ from dataclasses import dataclass -from typing import AsyncGenerator, Protocol, Sequence +from typing import Protocol, Sequence from autogen_core.base import CancellationToken from ..messages import ChatMessage, InnerMessage +from ._termination import TerminationCondition @dataclass @@ -22,16 +23,7 @@ async def run( task: str, *, cancellation_token: CancellationToken | None = None, + termination_condition: TerminationCondition | None = None, ) -> TaskResult: """Run the task.""" ... - - def run_stream( - self, - task: str, - *, - cancellation_token: CancellationToken | None = None, - ) -> AsyncGenerator[InnerMessage | ChatMessage | TaskResult, None]: - """Run the task and produces a stream of messages and the final result - :class:`TaskResult` as the last item in the stream.""" - ... diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_team.py b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_team.py index e112a3b512ed..b0a1dc3d2a38 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_team.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_team.py @@ -1,7 +1,18 @@ from typing import Protocol -from ._task import TaskRunner +from autogen_core.base import CancellationToken + +from ._task import TaskResult, TaskRunner +from ._termination import TerminationCondition class Team(TaskRunner, Protocol): - pass + async def run( + self, + task: str, + *, + cancellation_token: CancellationToken | None = None, + termination_condition: TerminationCondition | None = None, + ) -> TaskResult: + """Run the team on a given task until the termination condition is met.""" + ... diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_termination.py b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_termination.py index 1442dd51358a..0d0a056eab4c 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/base/_termination.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/base/_termination.py @@ -22,25 +22,19 @@ class TerminationCondition(ABC): .. code-block:: python - import asyncio from autogen_agentchat.teams import MaxTurnsTermination, TextMentionTermination + # Terminate the conversation after 10 turns or if the text "TERMINATE" is mentioned. + cond1 = MaxTurnsTermination(10) | TextMentionTermination("TERMINATE") - async def main() -> None: - # Terminate the conversation after 10 turns or if the text "TERMINATE" is mentioned. - cond1 = MaxTurnsTermination(10) | TextMentionTermination("TERMINATE") + # Terminate the conversation after 10 turns and if the text "TERMINATE" is mentioned. + cond2 = MaxTurnsTermination(10) & TextMentionTermination("TERMINATE") - # Terminate the conversation after 10 turns and if the text "TERMINATE" is mentioned. - cond2 = MaxTurnsTermination(10) & TextMentionTermination("TERMINATE") + ... - # ... - - # Reset the termination condition. - await cond1.reset() - await cond2.reset() - - - asyncio.run(main()) + # Reset the termination condition. + await cond1.reset() + await cond2.reset() """ @property diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/messages.py b/python/packages/autogen-agentchat/src/autogen_agentchat/messages.py index 1ac85edf1bd1..f206250e101e 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/messages.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/messages.py @@ -1,7 +1,7 @@ from typing import List from autogen_core.components import FunctionCall, Image -from autogen_core.components.models import FunctionExecutionResult, RequestUsage +from autogen_core.components.models import FunctionExecutionResult from pydantic import BaseModel @@ -11,9 +11,6 @@ class BaseMessage(BaseModel): source: str """The name of the agent that sent this message.""" - models_usage: RequestUsage | None = None - """The model client usage incurred when producing this message.""" - class TextMessage(BaseMessage): """A text message.""" @@ -60,14 +57,14 @@ class ToolCallMessage(BaseMessage): """The tool calls.""" -class ToolCallResultMessage(BaseMessage): +class ToolCallResultMessages(BaseMessage): """A message signaling the results of tool calls.""" content: List[FunctionExecutionResult] """The tool call results.""" -InnerMessage = ToolCallMessage | ToolCallResultMessage +InnerMessage = ToolCallMessage | ToolCallResultMessages """Messages for intra-agent monologues.""" @@ -83,6 +80,6 @@ class ToolCallResultMessage(BaseMessage): "HandoffMessage", "ResetMessage", "ToolCallMessage", - "ToolCallResultMessage", + "ToolCallResultMessages", "ChatMessage", ] diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/task/__init__.py b/python/packages/autogen-agentchat/src/autogen_agentchat/task/__init__.py index 0c1415d3bd1e..757edc043f38 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/task/__init__.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/task/__init__.py @@ -1,8 +1,7 @@ -from ._terminations import MaxMessageTermination, StopMessageTermination, TextMentionTermination, TokenUsageTermination +from ._terminations import MaxMessageTermination, StopMessageTermination, TextMentionTermination __all__ = [ "MaxMessageTermination", "TextMentionTermination", "StopMessageTermination", - "TokenUsageTermination", ] diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/task/_terminations.py b/python/packages/autogen-agentchat/src/autogen_agentchat/task/_terminations.py index 825d5bea28e6..ade11d759b36 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/task/_terminations.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/task/_terminations.py @@ -88,59 +88,3 @@ async def __call__(self, messages: Sequence[ChatMessage]) -> StopMessage | None: async def reset(self) -> None: self._terminated = False - - -class TokenUsageTermination(TerminationCondition): - """Terminate the conversation if a token usage limit is reached. - - Args: - max_total_token: The maximum total number of tokens allowed in the conversation. - max_prompt_token: The maximum number of prompt tokens allowed in the conversation. - max_completion_token: The maximum number of completion tokens allowed in the conversation. - - Raises: - ValueError: If none of max_total_token, max_prompt_token, or max_completion_token is provided. - """ - - def __init__( - self, - max_total_token: int | None = None, - max_prompt_token: int | None = None, - max_completion_token: int | None = None, - ) -> None: - if max_total_token is None and max_prompt_token is None and max_completion_token is None: - raise ValueError( - "At least one of max_total_token, max_prompt_token, or max_completion_token must be provided" - ) - self._max_total_token = max_total_token - self._max_prompt_token = max_prompt_token - self._max_completion_token = max_completion_token - self._total_token_count = 0 - self._prompt_token_count = 0 - self._completion_token_count = 0 - - @property - def terminated(self) -> bool: - return ( - (self._max_total_token is not None and self._total_token_count >= self._max_total_token) - or (self._max_prompt_token is not None and self._prompt_token_count >= self._max_prompt_token) - or (self._max_completion_token is not None and self._completion_token_count >= self._max_completion_token) - ) - - async def __call__(self, messages: Sequence[ChatMessage]) -> StopMessage | None: - if self.terminated: - raise TerminatedException("Termination condition has already been reached") - for message in messages: - if message.models_usage is not None: - self._prompt_token_count += message.models_usage.prompt_tokens - self._completion_token_count += message.models_usage.completion_tokens - self._total_token_count += message.models_usage.prompt_tokens + message.models_usage.completion_tokens - if self.terminated: - content = f"Token usage limit reached, total token count: {self._total_token_count}, prompt token count: {self._prompt_token_count}, completion token count: {self._completion_token_count}." - return StopMessage(content=content, source="TokenUsageTermination") - return None - - async def reset(self) -> None: - self._total_token_count = 0 - self._prompt_token_count = 0 - self._completion_token_count = 0 diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_base_group_chat.py b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_base_group_chat.py index cd4efe8f350e..9f3132a74955 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_base_group_chat.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_base_group_chat.py @@ -1,7 +1,6 @@ -import asyncio import uuid from abc import ABC, abstractmethod -from typing import AsyncGenerator, Callable, List +from typing import Callable, List from autogen_core.application import SingleThreadedAgentRuntime from autogen_core.base import ( @@ -74,25 +73,11 @@ async def run( task: str, *, cancellation_token: CancellationToken | None = None, + termination_condition: TerminationCondition | None = None, ) -> TaskResult: - """Run the team and return the result. The base implementation uses - :meth:`run_stream` to run the team and then returns the final result.""" - async for message in self.run_stream( - task, - cancellation_token=cancellation_token, - ): - if isinstance(message, TaskResult): - return message - raise AssertionError("The stream should have returned the final result.") - - async def run_stream( - self, - task: str, - *, - cancellation_token: CancellationToken | None = None, - ) -> AsyncGenerator[InnerMessage | ChatMessage | TaskResult, None]: - """Run the team and produces a stream of messages and the final result - of the type :class:`TaskResult` as the last item in the stream.""" + """Run the team and return the result.""" + # Create intervention handler for termination. + # Create the runtime. runtime = SingleThreadedAgentRuntime() @@ -132,7 +117,7 @@ async def run_stream( group_topic_type=group_topic_type, participant_topic_types=participant_topic_types, participant_descriptions=participant_descriptions, - termination_condition=self._termination_condition, + termination_condition=termination_condition or self._termination_condition, ), ) # Add subscriptions for the group chat manager. @@ -147,7 +132,6 @@ async def run_stream( ) output_messages: List[InnerMessage | ChatMessage] = [] - output_message_queue: asyncio.Queue[InnerMessage | ChatMessage | None] = asyncio.Queue() async def collect_output_messages( _runtime: AgentRuntime, @@ -156,7 +140,6 @@ async def collect_output_messages( ctx: MessageContext, ) -> None: output_messages.append(message) - await output_message_queue.put(message) await ClosureAgent.register( runtime, @@ -175,29 +158,14 @@ async def collect_output_messages( group_chat_manager_topic_id = TopicId(type=group_chat_manager_topic_type, source=self._team_id) first_chat_message = TextMessage(content=task, source="user") output_messages.append(first_chat_message) - await output_message_queue.put(first_chat_message) await runtime.publish_message( GroupChatPublishEvent(agent_message=first_chat_message), topic_id=team_topic_id, ) await runtime.publish_message(GroupChatRequestPublishEvent(), topic_id=group_chat_manager_topic_id) - # Start a coroutine to stop the runtime and signal the output message queue is complete. - async def stop_runtime() -> None: - await runtime.stop_when_idle() - await output_message_queue.put(None) - - shutdown_task = asyncio.create_task(stop_runtime()) - - # Yield the messsages until the queue is empty. - while True: - message = await output_message_queue.get() - if message is None: - break - yield message - - # Wait for the shutdown task to finish. - await shutdown_task + # Wait for the runtime to stop. + await runtime.stop_when_idle() - # Yield the final result. - yield TaskResult(messages=output_messages) + # Return the result. + return TaskResult(messages=output_messages) diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_chat_agent_container.py b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_chat_agent_container.py index 3fde3f6864b9..1423735c2f7c 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_chat_agent_container.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_chat_agent_container.py @@ -3,7 +3,7 @@ from autogen_core.base import MessageContext from autogen_core.components import DefaultTopicId, event -from ...base import ChatAgent, Response +from ...base import ChatAgent from ...messages import ChatMessage from .._events import GroupChatPublishEvent, GroupChatRequestPublishEvent from ._sequential_routed_agent import SequentialRoutedAgent @@ -37,26 +37,28 @@ async def handle_content_request(self, message: GroupChatRequestPublishEvent, ct """Handle a content request event by passing the messages in the buffer to the delegate agent and publish the response.""" # Pass the messages in the buffer to the delegate agent. - response: Response | None = None - async for msg in self._agent.on_messages_stream(self._message_buffer, ctx.cancellation_token): - if isinstance(msg, Response): - await self.publish_message( - msg.chat_message, - topic_id=DefaultTopicId(type=self._output_topic_type), - ) - response = msg - else: - # Publish the message to the output topic. - await self.publish_message(msg, topic_id=DefaultTopicId(type=self._output_topic_type)) - if response is None: - raise ValueError("The agent did not produce a final response. Check the agent's on_messages_stream method.") - - # Publish the response to the group chat. + response = await self._agent.on_messages(self._message_buffer, ctx.cancellation_token) + if not any(isinstance(response.chat_message, msg_type) for msg_type in self._agent.produced_message_types): + raise ValueError( + f"The agent {self._agent.name} produced an unexpected message type: {type(response)}. " + f"Expected one of: {self._agent.produced_message_types}. " + f"Check the agent's produced_message_types property." + ) + + # Publish inner messages to the output topic. + if response.inner_messages is not None: + for inner_message in response.inner_messages: + await self.publish_message(inner_message, topic_id=DefaultTopicId(type=self._output_topic_type)) + + # Publish the response. self._message_buffer.clear() await self.publish_message( GroupChatPublishEvent(agent_message=response.chat_message, source=self.id), topic_id=DefaultTopicId(type=self._parent_topic_type), ) + # Publish the response to the output topic. + await self.publish_message(response.chat_message, topic_id=DefaultTopicId(type=self._output_topic_type)) + async def on_unhandled_message(self, message: Any, ctx: MessageContext) -> None: raise ValueError(f"Unhandled message in agent container: {type(message)}") diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_round_robin_group_chat.py b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_round_robin_group_chat.py index 3ef4a2e07ad0..e8f5f66533f2 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_round_robin_group_chat.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_round_robin_group_chat.py @@ -50,8 +50,7 @@ class RoundRobinGroupChat(BaseGroupChat): Args: participants (List[BaseChatAgent]): The participants in the group chat. - termination_condition (TerminationCondition, optional): The termination condition for the group chat. Defaults to None. - Without a termination condition, the group chat will run indefinitely. + tools (List[Tool], optional): The tools to use in the group chat. Defaults to None. Raises: ValueError: If no participants are provided or if participant names are not unique. @@ -62,66 +61,32 @@ class RoundRobinGroupChat(BaseGroupChat): .. code-block:: python - import asyncio - from autogen_ext.models import OpenAIChatCompletionClient - from autogen_agentchat.agents import AssistantAgent - from autogen_agentchat.teams import RoundRobinGroupChat - from autogen_agentchat.task import TextMentionTermination + from autogen_agentchat.agents import ToolUseAssistantAgent + from autogen_agentchat.teams import RoundRobinGroupChat, StopMessageTermination - - async def main() -> None: - model_client = OpenAIChatCompletionClient(model="gpt-4o") - - async def get_weather(location: str) -> str: - return f"The weather in {location} is sunny." - - assistant = AssistantAgent( - "Assistant", - model_client=model_client, - tools=[get_weather], - ) - termination = TextMentionTermination("TERMINATE") - team = RoundRobinGroupChat([assistant], termination_condition=termination) - stream = team.run_stream("What's the weather in New York?") - async for message in stream: - print(message) - - - asyncio.run(main()) + assistant = ToolUseAssistantAgent("Assistant", model_client=..., registered_tools=...) + team = RoundRobinGroupChat([assistant]) + await team.run("What's the weather in New York?", termination_condition=StopMessageTermination()) A team with multiple participants: .. code-block:: python - import asyncio - from autogen_ext.models import OpenAIChatCompletionClient - from autogen_agentchat.agents import AssistantAgent - from autogen_agentchat.teams import RoundRobinGroupChat - from autogen_agentchat.task import TextMentionTermination - + from autogen_agentchat.agents import CodingAssistantAgent, CodeExecutorAgent + from autogen_agentchat.teams import RoundRobinGroupChat, StopMessageTermination - async def main() -> None: - model_client = OpenAIChatCompletionClient(model="gpt-4o") + coding_assistant = CodingAssistantAgent("Coding_Assistant", model_client=...) + executor_agent = CodeExecutorAgent("Code_Executor", code_executor=...) + team = RoundRobinGroupChat([coding_assistant, executor_agent]) + await team.run("Write a program that prints 'Hello, world!'", termination_condition=StopMessageTermination()) - agent1 = AssistantAgent("Assistant1", model_client=model_client) - agent2 = AssistantAgent("Assistant2", model_client=model_client) - termination = TextMentionTermination("TERMINATE") - team = RoundRobinGroupChat([agent1, agent2], termination_condition=termination) - stream = team.run_stream("Tell me some jokes.") - async for message in stream: - print(message) - - - asyncio.run(main()) """ - def __init__( - self, participants: List[ChatAgent], termination_condition: TerminationCondition | None = None - ) -> None: + def __init__(self, participants: List[ChatAgent], termination_condition: TerminationCondition | None = None): super().__init__( participants, - group_chat_manager_class=RoundRobinGroupChatManager, termination_condition=termination_condition, + group_chat_manager_class=RoundRobinGroupChatManager, ) def _create_group_chat_manager_factory( diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_selector_group_chat.py b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_selector_group_chat.py index 63b73cb88c2b..3cc489daa6b7 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_selector_group_chat.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_selector_group_chat.py @@ -1,12 +1,12 @@ import logging import re -from typing import Callable, Dict, List, Sequence +from typing import Callable, Dict, List from autogen_core.components.models import ChatCompletionClient, SystemMessage from ... import EVENT_LOGGER_NAME, TRACE_LOGGER_NAME from ...base import ChatAgent, TerminationCondition -from ...messages import ChatMessage, MultiModalMessage, StopMessage, TextMessage +from ...messages import MultiModalMessage, StopMessage, TextMessage from .._events import ( GroupChatPublishEvent, GroupChatSelectSpeakerEvent, @@ -20,7 +20,7 @@ class SelectorGroupChatManager(BaseGroupChatManager): """A group chat manager that selects the next speaker using a ChatCompletion - model and a custom selector function.""" + model.""" def __init__( self, @@ -32,7 +32,6 @@ def __init__( model_client: ChatCompletionClient, selector_prompt: str, allow_repeated_speaker: bool, - selector_func: Callable[[Sequence[ChatMessage]], str | None] | None, ) -> None: super().__init__( parent_topic_type, @@ -45,24 +44,12 @@ def __init__( self._selector_prompt = selector_prompt self._previous_speaker: str | None = None self._allow_repeated_speaker = allow_repeated_speaker - self._selector_func = selector_func async def select_speaker(self, thread: List[GroupChatPublishEvent]) -> str: - """Selects the next speaker in a group chat using a ChatCompletion client, - with the selector function as override if it returns a speaker name. + """Selects the next speaker in a group chat using a ChatCompletion client. A key assumption is that the agent type is the same as the topic type, which we use as the agent name. """ - - # Use the selector function if provided. - if self._selector_func is not None: - speaker = self._selector_func([msg.agent_message for msg in thread]) - if speaker is not None: - # Skip the model based selection. - event_logger.debug(GroupChatSelectSpeakerEvent(selected_speaker=speaker, source=self.id)) - return speaker - - # Construct the history of the conversation. history_messages: List[str] = [] for event in thread: msg = event.agent_message @@ -169,16 +156,10 @@ class SelectorGroupChat(BaseGroupChat): must have unique names and at least two participants. model_client (ChatCompletionClient): The ChatCompletion model client used to select the next speaker. - termination_condition (TerminationCondition, optional): The termination condition for the group chat. Defaults to None. - Without a termination condition, the group chat will run indefinitely. selector_prompt (str, optional): The prompt template to use for selecting the next speaker. Must contain '{roles}', '{participants}', and '{history}' to be filled in. allow_repeated_speaker (bool, optional): Whether to allow the same speaker to be selected consecutively. Defaults to False. - selector_func (Callable[[Sequence[ChatMessage]], str | None], optional): A custom selector - function that takes the conversation history and returns the name of the next speaker. - If provided, this function will be used to override the model to select the next speaker. - If the function returns None, the model will be used to select the next speaker. Raises: ValueError: If the number of participants is less than two or if the selector prompt is invalid. @@ -189,111 +170,14 @@ class SelectorGroupChat(BaseGroupChat): .. code-block:: python - import asyncio - from autogen_ext.models import OpenAIChatCompletionClient - from autogen_agentchat.agents import AssistantAgent - from autogen_agentchat.teams import SelectorGroupChat - from autogen_agentchat.task import TextMentionTermination - - - async def main() -> None: - model_client = OpenAIChatCompletionClient(model="gpt-4o") - - async def lookup_hotel(location: str) -> str: - return f"Here are some hotels in {location}: hotel1, hotel2, hotel3." - - async def lookup_flight(origin: str, destination: str) -> str: - return f"Here are some flights from {origin} to {destination}: flight1, flight2, flight3." - - async def book_trip() -> str: - return "Your trip is booked!" - - travel_advisor = AssistantAgent( - "Travel_Advisor", - model_client, - tools=[book_trip], - description="Helps with travel planning.", - ) - hotel_agent = AssistantAgent( - "Hotel_Agent", - model_client, - tools=[lookup_hotel], - description="Helps with hotel booking.", - ) - flight_agent = AssistantAgent( - "Flight_Agent", - model_client, - tools=[lookup_flight], - description="Helps with flight booking.", - ) - termination = TextMentionTermination("TERMINATE") - team = SelectorGroupChat( - [travel_advisor, hotel_agent, flight_agent], - model_client=model_client, - termination_condition=termination, - ) - stream = team.run_stream("Book a 3-day trip to new york.") - async for message in stream: - print(message) - - - asyncio.run(main()) - - A team with a custom selector function: - - .. code-block:: python - - import asyncio - from autogen_ext.models import OpenAIChatCompletionClient - from autogen_agentchat.agents import AssistantAgent - from autogen_agentchat.teams import SelectorGroupChat - from autogen_agentchat.task import TextMentionTermination - - - async def main() -> None: - model_client = OpenAIChatCompletionClient(model="gpt-4o") - - def check_caculation(x: int, y: int, answer: int) -> str: - if x + y == answer: - return "Correct!" - else: - return "Incorrect!" - - agent1 = AssistantAgent( - "Agent1", - model_client, - description="For calculation", - system_message="Calculate the sum of two numbers", - ) - agent2 = AssistantAgent( - "Agent2", - model_client, - tools=[check_caculation], - description="For checking calculation", - system_message="Check the answer and respond with 'Correct!' or 'Incorrect!'", - ) - - def selector_func(messages): - if len(messages) == 1 or messages[-1].content == "Incorrect!": - return "Agent1" - if messages[-1].source == "Agent1": - return "Agent2" - return None - - termination = TextMentionTermination("Correct!") - team = SelectorGroupChat( - [agent1, agent2], - model_client=model_client, - selector_func=selector_func, - termination_condition=termination, - ) - - stream = team.run_stream("What is 1 + 1?") - async for message in stream: - print(message) - + from autogen_agentchat.agents import ToolUseAssistantAgent + from autogen_agentchat.teams import SelectorGroupChat, StopMessageTermination - asyncio.run(main()) + travel_advisor = ToolUseAssistantAgent("Travel_Advisor", model_client=..., registered_tools=...) + hotel_agent = ToolUseAssistantAgent("Hotel_Agent", model_client=..., registered_tools=...) + flight_agent = ToolUseAssistantAgent("Flight_Agent", model_client=..., registered_tools=...) + team = SelectorGroupChat([travel_advisor, hotel_agent, flight_agent], model_client=...) + await team.run("Book a 3-day trip to new york.", termination_condition=StopMessageTermination()) """ def __init__( @@ -311,10 +195,9 @@ def __init__( Read the above conversation. Then select the next role from {participants} to play. Only return the role. """, allow_repeated_speaker: bool = False, - selector_func: Callable[[Sequence[ChatMessage]], str | None] | None = None, ): super().__init__( - participants, group_chat_manager_class=SelectorGroupChatManager, termination_condition=termination_condition + participants, termination_condition=termination_condition, group_chat_manager_class=SelectorGroupChatManager ) # Validate the participants. if len(participants) < 2: @@ -329,7 +212,6 @@ def __init__( self._selector_prompt = selector_prompt self._model_client = model_client self._allow_repeated_speaker = allow_repeated_speaker - self._selector_func = selector_func def _create_group_chat_manager_factory( self, @@ -348,5 +230,4 @@ def _create_group_chat_manager_factory( self._model_client, self._selector_prompt, self._allow_repeated_speaker, - self._selector_func, ) diff --git a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_swarm_group_chat.py b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_swarm_group_chat.py index 0a6bf9ee73fa..872f12e2ba31 100644 --- a/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_swarm_group_chat.py +++ b/python/packages/autogen-agentchat/src/autogen_agentchat/teams/_group_chat/_swarm_group_chat.py @@ -56,49 +56,35 @@ class Swarm(BaseGroupChat): Args: participants (List[ChatAgent]): The agents participating in the group chat. The first agent in the list is the initial speaker. - termination_condition (TerminationCondition, optional): The termination condition for the group chat. Defaults to None. - Without a termination condition, the group chat will run indefinitely. Examples: .. code-block:: python - import asyncio from autogen_ext.models import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import Swarm from autogen_agentchat.task import MaxMessageTermination + model_client = OpenAIChatCompletionClient(model="gpt-4o") - async def main() -> None: - model_client = OpenAIChatCompletionClient(model="gpt-4o") - - agent1 = AssistantAgent( - "Alice", - model_client=model_client, - handoffs=["Bob"], - system_message="You are Alice and you only answer questions about yourself.", - ) - agent2 = AssistantAgent( - "Bob", model_client=model_client, system_message="You are Bob and your birthday is on 1st January." - ) - - termination = MaxMessageTermination(3) - team = Swarm([agent1, agent2], termination_condition=termination) - - stream = team.run_stream("What is bob's birthday?") - async for message in stream: - print(message) - + agent1 = AssistantAgent( + "Alice", + model_client=model_client, + handoffs=["Bob"], + system_message="You are Alice and you only answer questions about yourself.", + ) + agent2 = AssistantAgent( + "Bob", model_client=model_client, system_message="You are Bob and your birthday is on 1st January." + ) - asyncio.run(main()) + team = Swarm([agent1, agent2]) + await team.run("What is bob's birthday?", termination_condition=MaxMessageTermination(3)) """ - def __init__( - self, participants: List[ChatAgent], termination_condition: TerminationCondition | None = None - ) -> None: + def __init__(self, participants: List[ChatAgent], termination_condition: TerminationCondition | None = None): super().__init__( - participants, group_chat_manager_class=SwarmGroupChatManager, termination_condition=termination_condition + participants, termination_condition=termination_condition, group_chat_manager_class=SwarmGroupChatManager ) # The first participant must be able to produce handoff messages. first_participant = self._participants[0] diff --git a/python/packages/autogen-agentchat/tests/test_assistant_agent.py b/python/packages/autogen-agentchat/tests/test_assistant_agent.py index 5e133b91a23d..9dee76539be4 100644 --- a/python/packages/autogen-agentchat/tests/test_assistant_agent.py +++ b/python/packages/autogen-agentchat/tests/test_assistant_agent.py @@ -6,9 +6,9 @@ import pytest from autogen_agentchat import EVENT_LOGGER_NAME from autogen_agentchat.agents import AssistantAgent, Handoff -from autogen_agentchat.base import TaskResult from autogen_agentchat.logging import FileLogHandler -from autogen_agentchat.messages import HandoffMessage, TextMessage, ToolCallMessage, ToolCallResultMessage +from autogen_agentchat.messages import HandoffMessage, TextMessage, ToolCallMessage, ToolCallResultMessages +from autogen_core.base import CancellationToken from autogen_core.components.tools import FunctionTool from autogen_ext.models import OpenAIChatCompletionClient from openai.resources.chat.completions import AsyncCompletions @@ -78,7 +78,7 @@ async def test_run_with_tools(monkeypatch: pytest.MonkeyPatch) -> None: created=0, model=model, object="chat.completion", - usage=CompletionUsage(prompt_tokens=10, completion_tokens=5, total_tokens=0), + usage=CompletionUsage(prompt_tokens=0, completion_tokens=0, total_tokens=0), ), ChatCompletion( id="id2", @@ -88,7 +88,7 @@ async def test_run_with_tools(monkeypatch: pytest.MonkeyPatch) -> None: created=0, model=model, object="chat.completion", - usage=CompletionUsage(prompt_tokens=10, completion_tokens=5, total_tokens=0), + usage=CompletionUsage(prompt_tokens=0, completion_tokens=0, total_tokens=0), ), ChatCompletion( id="id2", @@ -100,7 +100,7 @@ async def test_run_with_tools(monkeypatch: pytest.MonkeyPatch) -> None: created=0, model=model, object="chat.completion", - usage=CompletionUsage(prompt_tokens=10, completion_tokens=5, total_tokens=0), + usage=CompletionUsage(prompt_tokens=0, completion_tokens=0, total_tokens=0), ), ] mock = _MockChatCompletion(chat_completions) @@ -113,27 +113,9 @@ async def test_run_with_tools(monkeypatch: pytest.MonkeyPatch) -> None: result = await tool_use_agent.run("task") assert len(result.messages) == 4 assert isinstance(result.messages[0], TextMessage) - assert result.messages[0].models_usage is None assert isinstance(result.messages[1], ToolCallMessage) - assert result.messages[1].models_usage is not None - assert result.messages[1].models_usage.completion_tokens == 5 - assert result.messages[1].models_usage.prompt_tokens == 10 - assert isinstance(result.messages[2], ToolCallResultMessage) - assert result.messages[2].models_usage is None + assert isinstance(result.messages[2], ToolCallResultMessages) assert isinstance(result.messages[3], TextMessage) - assert result.messages[3].models_usage is not None - assert result.messages[3].models_usage.completion_tokens == 5 - assert result.messages[3].models_usage.prompt_tokens == 10 - - # Test streaming. - mock._curr_index = 0 # pyright: ignore - index = 0 - async for message in tool_use_agent.run_stream("task"): - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 @pytest.mark.asyncio @@ -166,7 +148,7 @@ async def test_handoffs(monkeypatch: pytest.MonkeyPatch) -> None: created=0, model=model, object="chat.completion", - usage=CompletionUsage(prompt_tokens=42, completion_tokens=43, total_tokens=85), + usage=CompletionUsage(prompt_tokens=0, completion_tokens=0, total_tokens=0), ), ] mock = _MockChatCompletion(chat_completions) @@ -178,27 +160,8 @@ async def test_handoffs(monkeypatch: pytest.MonkeyPatch) -> None: handoffs=[handoff], ) assert HandoffMessage in tool_use_agent.produced_message_types - result = await tool_use_agent.run("task") - assert len(result.messages) == 4 - assert isinstance(result.messages[0], TextMessage) - assert result.messages[0].models_usage is None - assert isinstance(result.messages[1], ToolCallMessage) - assert result.messages[1].models_usage is not None - assert result.messages[1].models_usage.completion_tokens == 43 - assert result.messages[1].models_usage.prompt_tokens == 42 - assert isinstance(result.messages[2], ToolCallResultMessage) - assert result.messages[2].models_usage is None - assert isinstance(result.messages[3], HandoffMessage) - assert result.messages[3].content == handoff.message - assert result.messages[3].target == handoff.target - assert result.messages[3].models_usage is None - - # Test streaming. - mock._curr_index = 0 # pyright: ignore - index = 0 - async for message in tool_use_agent.run_stream("task"): - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 + response = await tool_use_agent.on_messages( + [TextMessage(content="task", source="user")], cancellation_token=CancellationToken() + ) + assert isinstance(response.chat_message, HandoffMessage) + assert response.chat_message.target == "agent2" diff --git a/python/packages/autogen-agentchat/tests/test_group_chat.py b/python/packages/autogen-agentchat/tests/test_group_chat.py index 4922ceed9143..e6510c2fa17e 100644 --- a/python/packages/autogen-agentchat/tests/test_group_chat.py +++ b/python/packages/autogen-agentchat/tests/test_group_chat.py @@ -12,7 +12,7 @@ CodeExecutorAgent, Handoff, ) -from autogen_agentchat.base import Response, TaskResult +from autogen_agentchat.base import Response from autogen_agentchat.logging import FileLogHandler from autogen_agentchat.messages import ( ChatMessage, @@ -20,9 +20,9 @@ StopMessage, TextMessage, ToolCallMessage, - ToolCallResultMessage, + ToolCallResultMessages, ) -from autogen_agentchat.task import MaxMessageTermination, TextMentionTermination +from autogen_agentchat.task import MaxMessageTermination, StopMessageTermination from autogen_agentchat.teams import ( RoundRobinGroupChat, SelectorGroupChat, @@ -59,9 +59,6 @@ async def mock_create( self._curr_index += 1 return completion - def reset(self) -> None: - self._curr_index = 0 - class _EchoAgent(BaseChatAgent): def __init__(self, name: str, description: str) -> None: @@ -148,12 +145,9 @@ async def test_round_robin_group_chat(monkeypatch: pytest.MonkeyPatch) -> None: coding_assistant_agent = AssistantAgent( "coding_assistant", model_client=OpenAIChatCompletionClient(model=model, api_key="") ) - termination = TextMentionTermination("TERMINATE") - team = RoundRobinGroupChat( - participants=[coding_assistant_agent, code_executor_agent], termination_condition=termination - ) + team = RoundRobinGroupChat(participants=[coding_assistant_agent, code_executor_agent]) result = await team.run( - "Write a program that prints 'Hello, world!'", + "Write a program that prints 'Hello, world!'", termination_condition=StopMessageTermination() ) expected_messages = [ "Write a program that prints 'Hello, world!'", @@ -170,19 +164,6 @@ async def test_round_robin_group_chat(monkeypatch: pytest.MonkeyPatch) -> None: # Assert that all expected messages are in the collected messages assert normalized_messages == expected_messages - # Test streaming. - mock.reset() - index = 0 - await termination.reset() - async for message in team.run_stream( - "Write a program that prints 'Hello, world!'", - ): - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 - @pytest.mark.asyncio async def test_round_robin_group_chat_with_tools(monkeypatch: pytest.MonkeyPatch) -> None: @@ -247,19 +228,18 @@ async def test_round_robin_group_chat_with_tools(monkeypatch: pytest.MonkeyPatch tools=[tool], ) echo_agent = _EchoAgent("echo_agent", description="echo agent") - termination = TextMentionTermination("TERMINATE") - team = RoundRobinGroupChat(participants=[tool_use_agent, echo_agent], termination_condition=termination) + team = RoundRobinGroupChat(participants=[tool_use_agent, echo_agent]) result = await team.run( - "Write a program that prints 'Hello, world!'", + "Write a program that prints 'Hello, world!'", termination_condition=StopMessageTermination() ) assert len(result.messages) == 6 assert isinstance(result.messages[0], TextMessage) # task assert isinstance(result.messages[1], ToolCallMessage) # tool call - assert isinstance(result.messages[2], ToolCallResultMessage) # tool call result + assert isinstance(result.messages[2], ToolCallResultMessages) # tool call result assert isinstance(result.messages[3], TextMessage) # tool use agent response assert isinstance(result.messages[4], TextMessage) # echo agent response - assert isinstance(result.messages[5], TextMessage) # tool use agent response + assert isinstance(result.messages[5], StopMessage) # tool use agent response context = tool_use_agent._model_context # pyright: ignore assert context[0].content == "Write a program that prints 'Hello, world!'" @@ -273,20 +253,6 @@ async def test_round_robin_group_chat_with_tools(monkeypatch: pytest.MonkeyPatch assert context[2].content[0].call_id == "1" assert context[3].content == "Hello" - # Test streaming. - tool_use_agent._model_context.clear() # pyright: ignore - mock.reset() - index = 0 - await termination.reset() - async for message in team.run_stream( - "Write a program that prints 'Hello, world!'", - ): - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 - @pytest.mark.asyncio async def test_selector_group_chat(monkeypatch: pytest.MonkeyPatch) -> None: @@ -349,14 +315,12 @@ async def test_selector_group_chat(monkeypatch: pytest.MonkeyPatch) -> None: agent1 = _StopAgent("agent1", description="echo agent 1", stop_at=2) agent2 = _EchoAgent("agent2", description="echo agent 2") agent3 = _EchoAgent("agent3", description="echo agent 3") - termination = TextMentionTermination("TERMINATE") team = SelectorGroupChat( participants=[agent1, agent2, agent3], model_client=OpenAIChatCompletionClient(model=model, api_key=""), - termination_condition=termination, ) result = await team.run( - "Write a program that prints 'Hello, world!'", + "Write a program that prints 'Hello, world!'", termination_condition=StopMessageTermination() ) assert len(result.messages) == 6 assert result.messages[0].content == "Write a program that prints 'Hello, world!'" @@ -366,20 +330,6 @@ async def test_selector_group_chat(monkeypatch: pytest.MonkeyPatch) -> None: assert result.messages[4].source == "agent2" assert result.messages[5].source == "agent1" - # Test streaming. - mock.reset() - agent1._count = 0 # pyright: ignore - index = 0 - await termination.reset() - async for message in team.run_stream( - "Write a program that prints 'Hello, world!'", - ): - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 - @pytest.mark.asyncio async def test_selector_group_chat_two_speakers(monkeypatch: pytest.MonkeyPatch) -> None: @@ -401,14 +351,12 @@ async def test_selector_group_chat_two_speakers(monkeypatch: pytest.MonkeyPatch) agent1 = _StopAgent("agent1", description="echo agent 1", stop_at=2) agent2 = _EchoAgent("agent2", description="echo agent 2") - termination = TextMentionTermination("TERMINATE") team = SelectorGroupChat( participants=[agent1, agent2], - termination_condition=termination, model_client=OpenAIChatCompletionClient(model=model, api_key=""), ) result = await team.run( - "Write a program that prints 'Hello, world!'", + "Write a program that prints 'Hello, world!'", termination_condition=StopMessageTermination() ) assert len(result.messages) == 5 assert result.messages[0].content == "Write a program that prints 'Hello, world!'" @@ -419,18 +367,6 @@ async def test_selector_group_chat_two_speakers(monkeypatch: pytest.MonkeyPatch) # only one chat completion was called assert mock._curr_index == 1 # pyright: ignore - # Test streaming. - mock.reset() - agent1._count = 0 # pyright: ignore - index = 0 - await termination.reset() - async for message in team.run_stream("Write a program that prints 'Hello, world!'"): - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 - @pytest.mark.asyncio async def test_selector_group_chat_two_speakers_allow_repeated(monkeypatch: pytest.MonkeyPatch) -> None: @@ -472,81 +408,20 @@ async def test_selector_group_chat_two_speakers_allow_repeated(monkeypatch: pyte agent1 = _StopAgent("agent1", description="echo agent 1", stop_at=1) agent2 = _EchoAgent("agent2", description="echo agent 2") - termination = TextMentionTermination("TERMINATE") team = SelectorGroupChat( participants=[agent1, agent2], model_client=OpenAIChatCompletionClient(model=model, api_key=""), - termination_condition=termination, allow_repeated_speaker=True, ) - result = await team.run("Write a program that prints 'Hello, world!'") + result = await team.run( + "Write a program that prints 'Hello, world!'", termination_condition=StopMessageTermination() + ) assert len(result.messages) == 4 assert result.messages[0].content == "Write a program that prints 'Hello, world!'" assert result.messages[1].source == "agent2" assert result.messages[2].source == "agent2" assert result.messages[3].source == "agent1" - # Test streaming. - mock.reset() - index = 0 - await termination.reset() - async for message in team.run_stream("Write a program that prints 'Hello, world!'"): - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 - - -@pytest.mark.asyncio -async def test_selector_group_chat_custom_selector(monkeypatch: pytest.MonkeyPatch) -> None: - model = "gpt-4o-2024-05-13" - chat_completions = [ - ChatCompletion( - id="id2", - choices=[ - Choice(finish_reason="stop", index=0, message=ChatCompletionMessage(content="agent3", role="assistant")) - ], - created=0, - model=model, - object="chat.completion", - usage=CompletionUsage(prompt_tokens=0, completion_tokens=0, total_tokens=0), - ), - ] - mock = _MockChatCompletion(chat_completions) - monkeypatch.setattr(AsyncCompletions, "create", mock.mock_create) - agent1 = _EchoAgent("agent1", description="echo agent 1") - agent2 = _EchoAgent("agent2", description="echo agent 2") - agent3 = _EchoAgent("agent3", description="echo agent 3") - agent4 = _EchoAgent("agent4", description="echo agent 4") - - def _select_agent(messages: Sequence[ChatMessage]) -> str | None: - if len(messages) == 0: - return "agent1" - elif messages[-1].source == "agent1": - return "agent2" - elif messages[-1].source == "agent2": - return None - elif messages[-1].source == "agent3": - return "agent4" - else: - return "agent1" - - termination = MaxMessageTermination(6) - team = SelectorGroupChat( - participants=[agent1, agent2, agent3, agent4], - model_client=OpenAIChatCompletionClient(model=model, api_key=""), - selector_func=_select_agent, - termination_condition=termination, - ) - result = await team.run("task") - assert len(result.messages) == 6 - assert result.messages[1].source == "agent1" - assert result.messages[2].source == "agent2" - assert result.messages[3].source == "agent3" - assert result.messages[4].source == "agent4" - assert result.messages[5].source == "agent1" - class _HandOffAgent(BaseChatAgent): def __init__(self, name: str, description: str, next_agent: str) -> None: @@ -571,9 +446,8 @@ async def test_swarm_handoff() -> None: second_agent = _HandOffAgent("second_agent", description="second agent", next_agent="third_agent") third_agent = _HandOffAgent("third_agent", description="third agent", next_agent="first_agent") - termination = MaxMessageTermination(6) - team = Swarm([second_agent, first_agent, third_agent], termination_condition=termination) - result = await team.run("task") + team = Swarm([second_agent, first_agent, third_agent]) + result = await team.run("task", termination_condition=MaxMessageTermination(6)) assert len(result.messages) == 6 assert result.messages[0].content == "task" assert result.messages[1].content == "Transferred to third_agent." @@ -582,17 +456,6 @@ async def test_swarm_handoff() -> None: assert result.messages[4].content == "Transferred to third_agent." assert result.messages[5].content == "Transferred to first_agent." - # Test streaming. - index = 0 - await termination.reset() - stream = team.run_stream("task") - async for message in stream: - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 - @pytest.mark.asyncio async def test_swarm_handoff_using_tool_calls(monkeypatch: pytest.MonkeyPatch) -> None: @@ -651,33 +514,19 @@ async def test_swarm_handoff_using_tool_calls(monkeypatch: pytest.MonkeyPatch) - mock = _MockChatCompletion(chat_completions) monkeypatch.setattr(AsyncCompletions, "create", mock.mock_create) - agent1 = AssistantAgent( + agnet1 = AssistantAgent( "agent1", model_client=OpenAIChatCompletionClient(model=model, api_key=""), handoffs=[Handoff(target="agent2", name="handoff_to_agent2", message="handoff to agent2")], ) agent2 = _HandOffAgent("agent2", description="agent 2", next_agent="agent1") - termination = TextMentionTermination("TERMINATE") - team = Swarm([agent1, agent2], termination_condition=termination) - result = await team.run("task") + team = Swarm([agnet1, agent2]) + result = await team.run("task", termination_condition=StopMessageTermination()) assert len(result.messages) == 7 assert result.messages[0].content == "task" assert isinstance(result.messages[1], ToolCallMessage) - assert isinstance(result.messages[2], ToolCallResultMessage) + assert isinstance(result.messages[2], ToolCallResultMessages) assert result.messages[3].content == "handoff to agent2" assert result.messages[4].content == "Transferred to agent1." assert result.messages[5].content == "Hello" assert result.messages[6].content == "TERMINATE" - - # Test streaming. - agent1._model_context.clear() # pyright: ignore - mock.reset() - index = 0 - await termination.reset() - stream = team.run_stream("task") - async for message in stream: - if isinstance(message, TaskResult): - assert message == result - else: - assert message == result.messages[index] - index += 1 diff --git a/python/packages/autogen-agentchat/tests/test_termination_condition.py b/python/packages/autogen-agentchat/tests/test_termination_condition.py index 06362e156e01..7d504dce3a03 100644 --- a/python/packages/autogen-agentchat/tests/test_termination_condition.py +++ b/python/packages/autogen-agentchat/tests/test_termination_condition.py @@ -1,12 +1,6 @@ import pytest from autogen_agentchat.messages import StopMessage, TextMessage -from autogen_agentchat.task import ( - MaxMessageTermination, - StopMessageTermination, - TextMentionTermination, - TokenUsageTermination, -) -from autogen_core.components.models import RequestUsage +from autogen_agentchat.task import MaxMessageTermination, StopMessageTermination, TextMentionTermination @pytest.mark.asyncio @@ -57,51 +51,6 @@ async def test_mention_termination() -> None: ) -@pytest.mark.asyncio -async def test_token_usage_termination() -> None: - termination = TokenUsageTermination(max_total_token=10) - assert await termination([]) is None - await termination.reset() - assert ( - await termination( - [ - TextMessage( - content="Hello", source="user", models_usage=RequestUsage(prompt_tokens=10, completion_tokens=10) - ) - ] - ) - is not None - ) - await termination.reset() - assert ( - await termination( - [ - TextMessage( - content="Hello", source="user", models_usage=RequestUsage(prompt_tokens=1, completion_tokens=1) - ), - TextMessage( - content="World", source="agent", models_usage=RequestUsage(prompt_tokens=1, completion_tokens=1) - ), - ] - ) - is None - ) - await termination.reset() - assert ( - await termination( - [ - TextMessage( - content="Hello", source="user", models_usage=RequestUsage(prompt_tokens=5, completion_tokens=0) - ), - TextMessage( - content="stop", source="user", models_usage=RequestUsage(prompt_tokens=0, completion_tokens=5) - ), - ] - ) - is not None - ) - - @pytest.mark.asyncio async def test_and_termination() -> None: termination = MaxMessageTermination(2) & TextMentionTermination("stop") diff --git a/python/packages/autogen-core/docs/src/index.md b/python/packages/autogen-core/docs/src/index.md index 21a6fa35a922..4910721f53ab 100644 --- a/python/packages/autogen-core/docs/src/index.md +++ b/python/packages/autogen-core/docs/src/index.md @@ -61,7 +61,7 @@ AgentChat High-level API that includes preset agents and teams for building multi-agent systems. ```sh -pip install autogen-agentchat==0.4.0.dev3 +pip install autogen-agentchat==0.4.0.dev2 ``` 💡 *Start here if you are looking for an API similar to AutoGen 0.2* @@ -82,7 +82,7 @@ Get Started Provides building blocks for creating asynchronous, event driven multi-agent systems. ```sh -pip install autogen-core==0.4.0.dev3 +pip install autogen-core==0.4.0.dev2 ``` +++ diff --git a/python/packages/autogen-core/docs/src/packages/index.md b/python/packages/autogen-core/docs/src/packages/index.md index 7dd616108414..f471d4e48a3a 100644 --- a/python/packages/autogen-core/docs/src/packages/index.md +++ b/python/packages/autogen-core/docs/src/packages/index.md @@ -29,10 +29,10 @@ myst: Library that is at a similar level of abstraction as AutoGen 0.2, including default agents and group chat. ```sh -pip install autogen-agentchat==0.4.0.dev3 +pip install autogen-agentchat==0.4.0.dev2 ``` -[{fas}`circle-info;pst-color-primary` User Guide](/user-guide/agentchat-user-guide/index.md) | [{fas}`file-code;pst-color-primary` API Reference](/reference/python/autogen_agentchat/autogen_agentchat.rst) | [{fab}`python;pst-color-primary` PyPI](https://pypi.org/project/autogen-agentchat/0.4.0.dev3/) | [{fab}`github;pst-color-primary` Source](https://github.com/microsoft/autogen/tree/main/python/packages/autogen-agentchat) +[{fas}`circle-info;pst-color-primary` User Guide](/user-guide/agentchat-user-guide/index.md) | [{fas}`file-code;pst-color-primary` API Reference](/reference/python/autogen_agentchat/autogen_agentchat.rst) | [{fab}`python;pst-color-primary` PyPI](https://pypi.org/project/autogen-agentchat/0.4.0.dev2/) | [{fab}`github;pst-color-primary` Source](https://github.com/microsoft/autogen/tree/main/python/packages/autogen-agentchat) ::: (pkg-info-autogen-core)= @@ -44,10 +44,10 @@ pip install autogen-agentchat==0.4.0.dev3 Implements the core functionality of the AutoGen framework, providing basic building blocks for creating multi-agent systems. ```sh -pip install autogen-core==0.4.0.dev3 +pip install autogen-core==0.4.0.dev2 ``` -[{fas}`circle-info;pst-color-primary` User Guide](/user-guide/core-user-guide/index.md) | [{fas}`file-code;pst-color-primary` API Reference](/reference/python/autogen_core/autogen_core.rst) | [{fab}`python;pst-color-primary` PyPI](https://pypi.org/project/autogen-core/0.4.0.dev3/) | [{fab}`github;pst-color-primary` Source](https://github.com/microsoft/autogen/tree/main/python/packages/autogen-core) +[{fas}`circle-info;pst-color-primary` User Guide](/user-guide/core-user-guide/index.md) | [{fas}`file-code;pst-color-primary` API Reference](/reference/python/autogen_core/autogen_core.rst) | [{fab}`python;pst-color-primary` PyPI](https://pypi.org/project/autogen-core/0.4.0.dev2/) | [{fab}`github;pst-color-primary` Source](https://github.com/microsoft/autogen/tree/main/python/packages/autogen-core) ::: (pkg-info-autogen-ext)= @@ -59,7 +59,7 @@ pip install autogen-core==0.4.0.dev3 Implementations of core components that interface with external services, or use extra dependencies. For example, Docker based code execution. ```sh -pip install autogen-ext==0.4.0.dev3 +pip install autogen-ext==0.4.0.dev2 ``` Extras: @@ -69,7 +69,7 @@ Extras: - `docker` needed for {py:class}`~autogen_ext.code_executors.DockerCommandLineCodeExecutor` - `openai` needed for {py:class}`~autogen_ext.models.OpenAIChatCompletionClient` -[{fas}`circle-info;pst-color-primary` User Guide](/user-guide/extensions-user-guide/index.md) | [{fas}`file-code;pst-color-primary` API Reference](/reference/python/autogen_ext/autogen_ext.rst) | [{fab}`python;pst-color-primary` PyPI](https://pypi.org/project/autogen-ext/0.4.0.dev3/) | [{fab}`github;pst-color-primary` Source](https://github.com/microsoft/autogen/tree/main/python/packages/autogen-ext) +[{fas}`circle-info;pst-color-primary` User Guide](/user-guide/extensions-user-guide/index.md) | [{fas}`file-code;pst-color-primary` API Reference](/reference/python/autogen_ext/autogen_ext.rst) | [{fab}`python;pst-color-primary` PyPI](https://pypi.org/project/autogen-ext/0.4.0.dev2/) | [{fab}`github;pst-color-primary` Source](https://github.com/microsoft/autogen/tree/main/python/packages/autogen-ext) ::: (pkg-info-autogen-magentic-one)= diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/company-research.ipynb b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/company-research.ipynb index ce5844f4dec1..704abbf9b3de 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/company-research.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/company-research.ipynb @@ -18,12 +18,12 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "from autogen_agentchat.agents import CodingAssistantAgent, ToolUseAssistantAgent\n", - "from autogen_agentchat.task import TextMentionTermination\n", + "from autogen_agentchat.task import StopMessageTermination\n", "from autogen_agentchat.teams import RoundRobinGroupChat\n", "from autogen_core.components.tools import FunctionTool\n", "from autogen_ext.models import OpenAIChatCompletionClient" @@ -78,6 +78,22 @@ " if not api_key or not search_engine_id:\n", " raise ValueError(\"API key or Search Engine ID not found in environment variables\")\n", "\n", +"def google_search(query: str, num_results: int = 2, max_chars: int = 500) -> list: # type: ignore[type-arg]\n", + " import os\n", + " import time\n", + "\n", + " import requests\n", + " from bs4 import BeautifulSoup\n", + " from dotenv import load_dotenv\n", + "\n", + " load_dotenv()\n", + "\n", + " api_key = os.getenv(\"GOOGLE_API_KEY\")\n", + " search_engine_id = os.getenv(\"GOOGLE_SEARCH_ENGINE_ID\")\n", + "\n", + " if not api_key or not search_engine_id:\n", + " raise ValueError(\"API key or Search Engine ID not found in environment variables\")\n", + "\n", " url = \"https://www.googleapis.com/customsearch/v1\"\n", " params = {\"key\": str(api_key), \"cx\": str(search_engine_id), \"q\": str(query), \"num\": str(num_results)}\n", "\n", @@ -138,6 +154,21 @@ " return {\"error\": \"No historical data available for the specified ticker.\"}\n", "\n", " # Compute basic statistics and additional metrics\n", +" import yfinance as yf\n", + " from pytz import timezone # type: ignore\n", + "\n", + " stock = yf.Ticker(ticker)\n", + "\n", + " # Get historical data (1 year of data to ensure we have enough for 200-day MA)\n", + " end_date = datetime.now(timezone(\"UTC\"))\n", + " start_date = end_date - timedelta(days=365)\n", + " hist = stock.history(start=start_date, end=end_date)\n", + "\n", + " # Ensure we have data\n", + " if hist.empty:\n", + " return {\"error\": \"No historical data available for the specified ticker.\"}\n", + "\n", + " # Compute basic statistics and additional metrics\n", " current_price = stock.info.get(\"currentPrice\", hist[\"Close\"].iloc[-1])\n", " year_high = stock.info.get(\"fiftyTwoWeekHigh\", hist[\"High\"].max())\n", " year_low = stock.info.get(\"fiftyTwoWeekLow\", hist[\"Low\"].min())\n", @@ -191,6 +222,22 @@ "\n", " # Generate plot\n", " plt.figure(figsize=(12, 6))\n", +" \"52_week_low\": year_low,\n", + " \"50_day_ma\": ma_50,\n", + " \"200_day_ma\": ma_200,\n", + " \"ytd_price_change\": price_change,\n", + " \"ytd_percent_change\": percent_change,\n", + " \"trend\": trend,\n", + " \"volatility\": volatility,\n", + " }\n", + "\n", + " # Convert numpy types to Python native types for better JSON serialization\n", + " for key, value in result.items():\n", + " if isinstance(value, np.generic):\n", + " result[key] = value.item()\n", + "\n", + " # Generate plot\n", + " plt.figure(figsize=(12, 6))\n", " plt.plot(hist.index, hist[\"Close\"], label=\"Close Price\")\n", " plt.plot(hist.index, hist[\"Close\"].rolling(window=50).mean(), label=\"50-day MA\")\n", " plt.plot(hist.index, hist[\"Close\"].rolling(window=200).mean(), label=\"200-day MA\")\n", @@ -254,29 +301,39 @@ ")\n", "\n", "report_agent = CodingAssistantAgent(\n", +{ + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [], + "source": [ + "search_agent = ToolUseAssistantAgent(\n", + " name=\"Google_Search_Agent\",\n", + " model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\"),\n", + " registered_tools=[google_search_tool],\n", + " description=\"Search Google for information, returns top 2 results with a snippet and body content\",\n", + " system_message=\"You are a helpful AI assistant. Solve tasks using your tools.\",\n", + ")\n", + "\n", + "stock_analysis_agent = ToolUseAssistantAgent(\n", + " name=\"Stock_Analysis_Agent\",\n", + " model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\"),\n", + " registered_tools=[stock_analysis_tool],\n", + " description=\"Analyze stock data and generate a plot\",\n", + " system_message=\"You are a helpful AI assistant. Solve tasks using your tools.\",\n", + ")\n", + "\n", + "report_agent = CodingAssistantAgent(\n", " name=\"Report_Agent\",\n", " model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\"),\n", " description=\"Generate a report based on the search and stock analysis results\",\n", - " system_message=\"You are a helpful assistant that can generate a comprehensive report on a given topic based on search and stock analysis. When you done with generating the report, reply with TERMINATE.\",\n", + " system_message=\"You are a helpful assistant that can generate a comprehensive report on a given topic based on search and stock analysis. When you are done with generating the report, reply with TERMINATE.\",\n", ")\n", "\n", - "termination = TextMentionTermination(\"TERMINATE\")\n", - "team = RoundRobinGroupChat([search_agent, stock_analysis_agent, report_agent], termination_condition=termination)" - ] - }, - { - "cell_type": "code", - "execution_count": null, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "--------------------------------------------------------------------------- \n", - "\u001b[91m[2024-10-08T13:07:56.762630]:\u001b[0m\n", - "\n", + "team = RoundRobinGroupChat([search_agent, stock_analysis_agent, report_agent], termination_condition=TextMentionTermination(\"TERMINATE\"))" + ] +} +"\n", "Write a financial report on American airlines\n", "From: user\n", "--------------------------------------------------------------------------- \n", @@ -288,7 +345,7 @@ "\u001b[91m[2024-10-08T13:08:04.880251], tool_agent_for_Google_Search_Agent:\u001b[0m\n", "\n", "[FunctionExecutionResult(content=\"[{'title': 'American Airlines reports fourth-quarter and full-year 2023 financial ...', 'link': 'https://news.aa.com/news/news-details/2024/American-Airlines-reports-fourth-quarter-and-full-year-2023-financial-results-CORP-FI-01/default.aspx', 'snippet': 'Jan 25, 2024 ... American Airlines Group Inc. (NASDAQ: AAL) today reported its fourth-quarter and full-year 2023 financial results, including: Record\\\\xa0...', 'body': 'Just a moment... Enable JavaScript and cookies to continue'}, {'title': 'Financial AAL | American Airlines', 'link': 'https://americanairlines.gcs-web.com/financial-results/financial-aal', 'snippet': 'Financial Results. Investor Relations; Financial Results; Financial Results ... Transcript 196.4 KB. 2023. Q4. Earnings Release · Form 10-K · Transcript 190.6 KB\\\\xa0...', 'body': 'Financial AAL | American Airlines Skip to main navigation Main Menu Investor Relations Toolkit Presentations & Investor Updates Financial Results SEC Filings Annual Shareholders Meeting Proxy Materials & Virtual Shareholder Meeting AGM – QA Written Responses Stock Info Events Equity Distribution Analysts FAQs Merger Information Contact Us Corporate Information About Us Leadership Bios Fact Sheets Where We Fly Our Planes News Corporate Governance Plan Travel American Airlines AAdvantage'}, {'title': 'American Airlines Group - AnnualReports.com', 'link': 'https://www.annualreports.com/Company/american-airlines-group', 'snippet': 'Most Recent Annual Report · View 2023 Sustainability Report · Older/Archived Annual Reports · Rate This Report.', 'body': 'American Airlines Group - AnnualReports.com Menu BROWSE BY Exchanges Industry Other Filters BROWSE BY Exchanges Industry Other Filters 0 American Airlines Group Ticker AAL Exchange NASDAQ More Industry Air Services, Other More Sector Industrial Goods More 10,000+ Employees Based in Dallas-Fort Worth, Texas American Airlines offers customers 6,800 daily flights to more than 365 destinations in 61 countries from its hubs in Charlotte, Chicago, Dallas-Fort Worth, Los Angeles, Miami, New York,'}, {'title': 'American Airlines reports fourth-quarter and full-year 2023 financial ...', 'link': 'https://americanairlines.gcs-web.com/news-releases/news-release-details/american-airlines-reports-fourth-quarter-and-full-year-2023', 'snippet': 'Jan 25, 2024 ... American produced record revenue of nearly $53 billion. In the fourth quarter, the company generated revenue of more than $13 billion and an operating margin\\\\xa0...', 'body': 'American Airlines reports fourth-quarter and full-year 2023 financial results | American Airlines Skip to main navigation Main Menu Investor Relations Toolkit Presentations & Investor Updates Financial Results SEC Filings Annual Shareholders Meeting Proxy Materials & Virtual Shareholder Meeting AGM – QA Written Responses Stock Info Events Equity Distribution Analysts FAQs Merger Information Contact Us Corporate Information About Us Leadership Bios Fact Sheets Where We Fly Our Planes News'}, {'title': 'American Airlines reports fourth-quarter and full-year 2023 financial ...', 'link': 'https://finance.yahoo.com/news/american-airlines-reports-fourth-quarter-120000360.html', 'snippet': 'Jan 25, 2024 ... FORT WORTH, Texas, Jan. 25, 2024 (GLOBE NEWSWIRE) -- American Airlines Group Inc. (NASDAQ: AAL) today reported its fourth-quarter and\\\\xa0...', 'body': 'Yahoo Will be right back... Thank you for your patience. Our engineers are working quickly to resolve the issue.'}]\", call_id='call_Q72sGSLMyu2CLa7kVzZDHMpL')]\n", - "From: tool_agent_for_Google_Search_Agent\n", +"From: tool_agent_for_Google_Search_Agent\n", "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-08T13:08:06.643034], Google_Search_Agent:\u001b[0m\n", "\n", @@ -305,6 +362,13 @@ "### Financial Report: American Airlines Group Inc. (2023)\n", "\n", "#### Overview\n", +"From: tool_agent_for_Google_Search_Agent\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-08T13:08:16.385923], Google_Search_Agent:\u001b[0m\n", + "\n", + "### Financial Report: American Airlines Group Inc. (2023)\n", + "\n", + "#### Overview\n", "American Airlines Group Inc. (NASDAQ: AAL), one of the largest airlines in the world, released its financial results for the fourth quarter and full year of 2023 on January 25, 2024.\n", "\n", "#### Key Financial Highlights for Full Year 2023\n", @@ -343,6 +407,10 @@ "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-08T13:08:28.314277], Stock_Analysis_Agent:\u001b[0m\n", "\n", +"From: tool_agent_for_Stock_Analysis_Agent\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-08T13:08:28.314277], Stock_Analysis_Agent:\u001b[0m\n", + "\n", "### Financial Report: American Airlines Group Inc. (2023)\n", "\n", "#### Overview\n", @@ -386,13 +454,136 @@ "\u001b[91m[2024-10-08T13:08:28.703160], Report_Agent:\u001b[0m\n", "\n", "TERMINATE\n", - "From: Report_AgentTeamRunResult(messages=[TextMessage(source='user', content='Write a financial report on American airlines'), TextMessage(source='Google_Search_Agent', content=\"### Financial Report: American Airlines Group Inc. (2023)\\n\\n#### Overview\\nAmerican Airlines Group Inc. (NASDAQ: AAL), one of the largest airlines in the world, released its financial results for the fourth quarter and full year of 2023 on January 25, 2024.\\n\\n#### Key Financial Highlights for Full Year 2023\\n- **Total Revenue**: American Airlines achieved record revenue of nearly **$53 billion** for the full year.\\n- **Operating Revenue**: In the fourth quarter, the airline generated more than **$13 billion** in revenue.\\n- **Operating Margin**: The operating margin improved significantly due to increased demand and operational efficiencies.\\n\\n#### Q4 Financial Results\\n- **Revenue in Q4**: The fourth quarter witnessed a robust revenue growth that reflects strong travel demand.\\n- **Performance Metrics**: Additional performance metrics include:\\n - Detailed information about passenger revenues, cargo revenues, and operational metrics such as capacity and load factors.\\n \\n#### Future Outlook\\nAmerican Airlines is focusing on expanding its flight services and improving overall operational efficiency in 2024. The airline aims to capitalize on the continuing recovery in travel demand, which should positively impact future revenues and profitability.\\n\\n#### Corporate Strategy\\nThe airline has emphasized the importance of sustainable practices and operational innovations in its corporate strategy to adapt to evolving market conditions.\\n\\n#### Links for More Information\\n- [American Airlines Q4 2023 Financial Results](https://news.aa.com/news/news-details/2024/American-Airlines-reports-fourth-quarter-and-full-year-2023-financial-results-CORP-FI-01/default.aspx)\\n- [Investor Relations Page](http://americanairlines.gcs-web.com/)\\n\\nThis financial report provides a summary of American Airlines' financial performance and outlook as of 2023. For detailed insights, the complete financial statements and press releases should be consulted directly from the airline's investor relations resources.\"), TextMessage(source='Stock_Analysis_Agent', content=\"### Financial Report: American Airlines Group Inc. (2023)\\n\\n#### Overview\\nAmerican Airlines Group Inc. (NASDAQ: AAL) is one of the largest airlines globally, with substantial operations in passenger and cargo transport. The company has recently released its financial results for the fourth quarter and the full year of 2023.\\n\\n#### Key Financial Highlights for Full Year 2023\\n- **Total Revenue**: American Airlines achieved record revenue of nearly **$53 billion** for the full year.\\n- **Operating Revenue**: The airline generated over **$13 billion** in revenue during the fourth quarter.\\n- **Operating Margin**: There was a significant improvement in the operating margin, driven by increased travel demand and operational efficiencies.\\n\\n#### Q4 Financial Results\\n- The fourth quarter showed robust revenue growth, reflecting strong demand for travel.\\n- Additional performance metrics included passenger revenues, cargo revenues, capacity, and load factors, showcasing the airline's recovery post-pandemic.\\n\\n#### Stock Performance Overview\\n- **Current Stock Price**: $11.92\\n- **52-Week High**: $16.15\\n- **52-Week Low**: $9.07\\n- **50-Day Moving Average**: $11.84\\n- **200-Day Moving Average**: $12.71\\n- **Year-to-Date Price Change**: -$1.52\\n- **Year-to-Date Percent Change**: -11.31%\\n- **Trend**: The stock is currently in a downward trend.\\n- **Volatility**: The stock has a volatility of 41.10%.\\n\\n![American Airlines Stock Price Trend](coding/AAL_stockprice.png)\\n\\n#### Future Outlook\\nAmerican Airlines aims to expand its services and improve operational efficiency throughout 2024. The airline is poised to benefit from the continuing recovery in travel demand, which is expected to positively affect future revenues and profitability.\\n\\n#### Corporate Strategy\\nThe airline has prioritized sustainability and innovation within its operational strategy to adapt to ever-changing market conditions.\\n\\n#### Links for More Information\\n- [American Airlines Q4 2023 Financial Results](https://news.aa.com/news/news-details/2024/American-Airlines-reports-fourth-quarter-and-full-year-2023-financial-results-CORP-FI-01/default.aspx)\\n- [Investor Relations Page](http://americanairlines.gcs-web.com/)\\n\\nThis financial report provides a comprehensive overview of American Airlines' financial performance and stock analysis as of 2023. For more detailed insights, please refer to the airline's investor relations resources.\"), StopMessage(source='Report_Agent', content='TERMINATE')])\n" - ] +"From: Report_AgentTeamRunResult(messages=[TextMessage(source='user', content='Write a financial report on American airlines'), TextMessage(source='Google_Search_Agent', content=\"### Financial Report: American Airlines Group Inc. (2023)\\n\\n#### Overview\\nAmerican Airlines Group Inc. (NASDAQ: AAL), one of the largest airlines in the world, released its financial results for the fourth quarter and full year of 2023 on January 25, 2024.\\n\\n#### Key Financial Highlights for Full Year 2023\\n- **Total Revenue**: American Airlines achieved record revenue of nearly **$53 billion** for the full year.\\n- **Operating Revenue**: In the fourth quarter, the airline generated more than **$13 billion** in revenue.\\n- **Operating Margin**: The operating margin improved significantly due to increased demand and operational efficiencies.\\n\\n#### Q4 Financial Results\\n- **Revenue in Q4**: The fourth quarter witnessed a robust revenue growth that reflects strong travel demand.\\n- **Performance Metrics**: Additional performance metrics include:\\n - Detailed information about passenger revenues, cargo revenues, and operational metrics such as capacity and load factors.\\n \\n#### Future Outlook\\nAmerican Airlines is focusing on expanding its flight services and improving overall operational efficiency in 2024. The airline aims to capitalize on the continuing recovery in travel demand, which should positively impact future revenues and profitability.\\n\\n#### Corporate Strategy\\nThe airline has emphasized the importance of sustainable practices and operational innovations in its corporate strategy to adapt to evolving market conditions.\\n\\n#### Links for More Information\\n- [American Airlines Q4 2023 Financial Results](https://news.aa.com/news/news-details/2024/American-Airlines-reports-fourth-quarter-and-full-year-2023-financial-results-CORP-FI-01/default.aspx)\\n- [Investor Relations Page](http://americanairlines.gcs-web.com/)\\n\\nThis financial report provides a summary of American Airlines' financial performance and outlook as of 2023. For detailed insights, the complete financial statements and press releases should be consulted directly from the airline's investor relations resources.\"), TextMessage(source='Stock_Analysis_Agent', content=\"### Financial Report: American Airlines Group Inc. (2023)\\n\\n#### Overview\\nAmerican Airlines Group Inc. (NASDAQ: AAL) is one of the largest airlines globally, with substantial operations in passenger and cargo transport. The company has recently released its financial results for the fourth quarter and the full year of 2023.\\n\\n#### Key Financial Highlights for Full Year 2023\\n- **Total Revenue**: American Airlines achieved record revenue of nearly **$53 billion** for the full year.\\n- **Operating Revenue**: The airline generated over **$13 billion** in revenue during the fourth quarter.\\n- **Operating Margin**: There was a significant improvement in the operating margin, driven by increased travel demand and operational efficiencies.\\n\\n#### Q4 Financial Results\\n- The fourth quarter showed robust revenue growth, reflecting strong demand for travel.\\n- Additional performance metrics included passenger revenues, cargo revenues, capacity, and load factors, showcasing the airline's recovery post-pandemic.\\n\\n#### Stock Performance Overview\\n- **Current Stock Price**: $11.92\\n- **52-Week High**: $16.15\\n- **52-Week Low**: $9.07\\n- **50-Day Moving Average**: $11.84\\n- **200-Day Moving Average**: $12.71\\n- **Year-to-Date Price Change**: -$1.52\\n- **Year-to-Date Percent Change**: -11.31%\\n- **Trend**: The stock is currently in a downward trend.\\n- **Volatility**: The stock has a volatility of 41.10%.\\n\\n![American Airlines Stock Price Trend](coding/AAL_stockprice.png)\\n\\n#### Future Outlook\\nAmerican Airlines aims to expand its services and improve operational efficiency throughout 2024. The airline is poised to benefit from the continuing recovery in travel demand, which is expected to positively affect future revenues and profitability.\\n\\n#### Corporate Strategy\\nThe airline has prioritized sustainability and innovation within its operational strategy to adapt to ever-changing market conditions.\\n\\n#### Links for More Information\\n- [American Airlines Q4 2023 Financial Results](https://news.aa.com/news/news-details/2024/American-Airlines-reports-fourth-quarter-and-full-year-2023-financial-results-CORP-FI-01/default.aspx)\\n- [Investor Relations +**Trend**: The stock is currently in a downward trend.\\n- **Volatility**: The stock has a volatility of 41.10%.\\n\\n![American Airlines Stock Price Trend](coding/AAL_stockprice.png)\\n\\n#### Future Outlook\\nAmerican Airlines aims to expand its services and improve operational efficiency throughout 2024. The airline is poised to benefit from the continuing recovery in travel demand, which is expected to positively affect future revenues and profitability.\\n\\n#### Corporate Strategy\\nThe airline has prioritized sustainability and innovation within its operational strategy to adapt to ever-changing market conditions.\\n\\n#### Links for More Information\\n- [American Airlines Q4 2023 Financial Results](https://news.aa.com/news/news-details/2024/American-Airlines-reports-fourth-quarter-and-full-year-2023-financial-results-CORP-FI-01/default.aspx)\\n- [Investor Relations Page](http://americanairlines.gcs-web.com/)\\n\\nThis financial report provides a comprehensive overview of American Airlines' financial performance and stock analysis as of 2023. For more detailed insights, please refer to the airline's investor relations resources.\"), StopMessage(source='Report_Agent', content='TERMINATE')])\n" +] }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAA+UAAAIjCAYAAABlBbqXAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8hTgPZAAAACXBIWXMAAA9hAAAPYQGoP6dpAAEAAElEQVR4nOzdd3hUZfbA8e+09J6QhEAg9NBBOqgUQUREURTFhmJbRV17+9lYdV3X3lZ3XYoF66rYEAQUQboovUNIIKT3Pu3+/rhzJwlpkzYzCefzPHlIMndm3plMwpx7znuOTlEUBSGEEEIIIYQQQrid3tMLEEIIIYQQQgghzlQSlAshhBBCCCGEEB4iQbkQQgghhBBCCOEhEpQLIYQQQgghhBAeIkG5EEIIIYQQQgjhIRKUCyGEEEIIIYQQHiJBuRBCCCGEEEII4SESlAshhBBCCCGEEB4iQbkQQgghhBBCCOEhEpQLIYQQXuTpp59Gp9ORnZ3tkfu/4YYbSEhI8Mh9//Of/yQxMRG73e6R+2/LLBYL8fHx/Otf//L0UoQQQjSSBOVCCCHc7l//+hc6nY5Ro0a5dPxDDz2ETqfjyiuvrPXy48ePo9PpeOmll5q0nu+++47x48cTHR1NQEAA3bt3Z/bs2axYscJ5zKlTp3j66afZsWNHk+7DHZYsWYJOp3N++Pn50bt3b+68804yMjI8vbx6FRYW8sILL/Dwww+j11e+Pan6ePR6PXFxcZx//vmsXbu2Vdbxr3/9iyVLljR43KeffopOp+Pf//53rZfffvvtmEwmdu7c2cIrrJ3JZOK+++7jueeeo7y83C33KYQQomVIUC6EEMLtli5dSkJCAlu3buXIkSP1HqsoCp988gkJCQl89913FBUVtehaXnrpJS6++GJ0Oh2PPvoor776KrNmzeLw4cN8+umnzuNOnTrFggULvDoo1/ztb3/jww8/5K233mLs2LG88847jBkzhtLS0gav+95773Hw4EE3rLK6RYsWYbVamTNnTo3LpkyZwocffsj777/PX/7yF3bt2sWkSZP48ccfW3wdrgblV111FRdccAGPPPJIjRMeW7du5T//+Q/33nsvgwcPbvE11uXGG28kOzubjz/+2G33KYQQovmMnl6AEEKIM0tSUhIbN27kq6++4rbbbmPp0qU89dRTdR6/du1aTp48yc8//8zUqVP56quvmDt3bousxWq18swzzzBlyhR++umnGpdnZma2yP2427Rp0xg+fDgAN998M5GRkbzyyit88803tQa9ACUlJQQGBmIymdy5VKfFixdz8cUX4+fnV+Oy3r17c+211zq/vvTSSxk0aBCvvfYa06ZNc+cyq3nnnXfo378/9957rzMQttls3HbbbXTp0oWnn3661degKArl5eX4+/sTFhbG+eefz5IlS5g3b16r37cQQoiWIZlyIYQQbrV06VLCw8OZPn06l19+OUuXLm3w+H79+jFx4kQmT57c4PGNkZ2dTWFhIePGjav18ujoaEA9MTBixAhAzUZq5dRVM6pffPEFw4YNw9/fn6ioKK699lpSU1Nr3OaBAweYPXs2HTp0wN/fnz59+vB///d/9a4zOTmZnj17MmDAgCaVoU+aNAlQT4iAum88KCiIo0ePcuGFFxIcHMw111zjvOz0PeV2u53XX3+dgQMH4ufnR4cOHbjgggv4/fffqx330UcfOZ+DiIgIrrrqKk6cONHg+pKSkti1axeTJ0926fEMHDiQqKgo5+NZv349V1xxBV26dMHX15f4+HjuvfdeysrKql0vPT2dG2+8kc6dO+Pr60vHjh255JJLOH78OAAJCQns3buXX3/91fkznjBhQp3rSEhI4Omnn+aTTz5h1apVALzxxhvs2LGDd955h4CAACoqKnjqqafo2bOnc20PPfQQFRUV1W5r8eLFTJo0iejoaHx9fenXrx/vvPNOrfd50UUXsXLlSoYPH46/v3+1EvopU6bw22+/kZub69JzKYQQwvMkKBdCCOFWS5cu5bLLLsPHx4c5c+Zw+PBhtm3bVuuxFRUVfPnll87s7pw5c/j5559JT09vkbVER0fj7+/Pd999V28Q07dvX/72t78BcOutt/Lhhx/y4Ycfcu655wLqXu7Zs2djMBh4/vnnueWWW/jqq684++yzyc/Pd97Orl27GDVqFD///DO33HILr7/+OjNnzuS7776r876PHj3KueeeS3BwMGvXriUmJqbRj/Po0aMAREZGOr9ntVqZOnUq0dHRvPTSS8yaNavO6990003cc889xMfH88ILL/DII4/g5+fH5s2bncc899xzXH/99fTq1YtXXnmFe+65hzVr1nDuuedWew5qs3HjRgDOOusslx5PXl4eeXl5zsfzxRdfUFpayu23386bb77J1KlTefPNN7n++uurXW/WrFl8/fXX3HjjjfzrX//i7rvvpqioiJSUFABee+01OnfuTGJiovNn3NAJE61E/fbbb+fIkSM8+eSTztJ2u93OxRdfzEsvvcSMGTN48803mTlzJq+++mqN/gjvvPMOXbt25bHHHuPll18mPj6eO+64g7fffrvGfR48eJA5c+YwZcoUXn/9dYYMGeK8bNiwYSiK4nxOhRBCtAGKEEII4Sa///67AiirVq1SFEVR7Ha70rlzZ+Wvf/1rrcf/73//UwDl8OHDiqIoSmFhoeLn56e8+uqr1Y5LSkpSAOXFF19s9JqefPJJBVACAwOVadOmKc8995yyffv2Gsdt27ZNAZTFixdX+77ZbFaio6OVAQMGKGVlZc7vf//99wqgPPnkk87vnXvuuUpwcLCSnJxc7Tbsdrvz86eeekoBlKysLGX//v1KXFycMmLECCU3N7fBx7J48WIFUFavXq1kZWUpJ06cUD799FMlMjJS8ff3V06ePKkoiqLMnTtXAZRHHnmkxm3MnTtX6dq1q/Prn3/+WQGUu+++u8ax2rqPHz+uGAwG5bnnnqt2+e7duxWj0Vjj+6d7/PHHFUApKiqqcRmg3HTTTUpWVpaSmZmpbNmyRTnvvPMUQHn55ZcVRVGU0tLSGtd7/vnnFZ1O53yu8/LyXHqN9O/fXxk/fny9x5xuy5Ytil6vVyIiIpSwsDAlPT1dURRF+fDDDxW9Xq+sX7++2vHvvvuuAigbNmxwfq+2xzB16lSle/fu1b7XtWtXBVBWrFhR61pOnTqlAMoLL7zQqMcghBDCcyRTLoQQwm2WLl1KTEwMEydOBHB2VP/000+x2Wy1Hj98+HB69uwJQHBwMNOnT2/REvYFCxbw8ccfM3ToUFauXMn//d//MWzYMM466yz279/f4PV///13MjMzueOOO6rth54+fTqJiYn88MMPAGRlZbFu3TrmzZtHly5dqt2GTqercbt79uxh/PjxJCQksHr1asLDw11+TJMnT6ZDhw7Ex8dz1VVXERQUxNdff02nTp2qHXf77bc3eFtffvklOp2u1n3/2rq/+uor7HY7s2fPJjs72/kRGxtLr169+OWXX+q9j5ycHIxGI0FBQbVevnDhQjp06EB0dDSjRo1iw4YN3Hfffdxzzz0A+Pv7O48tKSkhOzubsWPHoigKf/75p/MYHx8f1q5dS15eXoOPuzFGjhzJX/7yF3Jzc3n++eed1QxffPEFffv2JTExsdrzom0nqPq8VH0MBQUFZGdnM378eI4dO0ZBQUG1++vWrRtTp06tdS3a68RTI/WEEEI0njR6E0II4RY2m41PP/2UiRMnOvcCA4waNYqXX36ZNWvWcP755zu/n5+fz/Lly7nzzjurdWgfN24cX375JYcOHaJ3794tsrY5c+YwZ84cCgsL2bJlC0uWLOHjjz9mxowZ7Nmzp9bmY5rk5GQA+vTpU+OyxMREfvvtNwCOHTsGwIABA1xa04wZM4iJiWHlypV1Bqt1efvtt+nduzdGo5GYmBj69OlTbcwYgNFopHPnzg3e1tGjR4mLiyMiIqLOYw4fPoyiKPTq1avWy5vbPO6SSy7hzjvvRKfTERwcTP/+/QkMDHRenpKSwpNPPsm3335bI+DWAlpfX19eeOEF7r//fmJiYhg9ejQXXXQR119/PbGxsc1aH+DsOaA12AP1edm/fz8dOnSo9TpVGwlu2LCBp556ik2bNtXokl9QUEBoaKjz627dutW5DkVRgNpP9AghhPBOEpQLIYRwi59//pm0tDQ+/fTTaqPGNEuXLq0WlH/xxRdUVFTw8ssv8/LLL9d6/IIFC1p0jSEhIUyZMoUpU6ZgMpl4//332bJlC+PHj2/R+3HFrFmzeP/991m6dCm33XZbo647cuTIasFhbXx9fWsE6k1lt9vR6XT8+OOPGAyGGpc3dFIhMjISq9VKUVERwcHBNS7v3LlznU3gbDYbU6ZMITc3l4cffpjExEQCAwNJTU3lhhtuwG63O4+95557mDFjBsuWLWPlypU88cQTPP/88/z8888MHTq0kY+6YXa7nYEDB/LKK6/Uenl8fDygnvg477zzSExM5JVXXiE+Ph4fHx+WL1/Oq6++Wu0xQPWs+um0kxJRUVEt9CiEEEK0NgnKhRBCuMXSpUuJjo6utXHVV199xddff827777rDDiWLl3KgAEDai2b/ve//83HH3/c4kF5VcOHD+f9998nLS0NqDvz2LVrV0BtvqWVJWsOHjzovLx79+6AWpbuihdffBGj0cgdd9xBcHAwV199dZMeR3P16NGDlStXkpubW2e2vEePHiiKQrdu3ZpUvZCYmAioXdgHDRrUqOvu3r2bQ4cO8f7771dr7KZ1Q69trffffz/3338/hw8fZsiQIbz88st89NFHQMtmmHv06MHOnTs577zz6r3d7777joqKCr799ttqWxsaKvuvjVaF0rdv38YvWAghhEfInnIhhBCtrqysjK+++oqLLrqIyy+/vMbHnXfeSVFREd9++y0AJ06cYN26dcyePbvW42+88UaOHDnCli1bmrWu0tJSNm3aVOtlP/74I1BZlq6VS5/eSXz48OFER0fz7rvvVhtz9eOPP7J//36mT58OQIcOHTj33HNZtGiRs9u3Ris5rkqn0/Gf//yHyy+/nLlz5zqfG3ebNWsWiqLUegJEW/dll12GwWBgwYIFNR6Loijk5OTUex9jxowBqDFizRVaZr7q/SqKwuuvv17tuNLSUsrLy6t9r0ePHgQHB1f7uQUGBjbYLd5Vs2fPJjU1lffee6/GZWVlZZSUlNT5GAoKCli8eHGj73P79u3odDrncyqEEML7SaZcCCFEq/v2228pKiri4osvrvXy0aNH06FDB5YuXcqVV17Jxx9/jKIodR5/4YUXYjQaWbp0KaNGjXJ+f82aNTUCL4CZM2fWupe7tLSUsWPHMnr0aC644ALi4+PJz89n2bJlrF+/npkzZzrLmnv06EFYWBjvvvsuwcHBBAYGMmrUKLp168YLL7zAjTfeyPjx45kzZw4ZGRm8/vrrJCQkcO+99zrv74033uDss8/mrLPO4tZbb6Vbt24cP36cH374gR07dtRYn16v56OPPmLmzJnMnj2b5cuX18jGt7aJEydy3XXX8cYbb3D48GHnqK/169czceJE7rzzTnr06MGzzz7Lo48+yvHjx5k5cybBwcEkJSXx9ddfc+utt/LAAw/UeR/du3dnwIABrF69mnnz5jVqfYmJifTo0YMHHniA1NRUQkJC+PLLL2vsLT906BDnnXces2fPpl+/fhiNRr7++msyMjK46qqrnMcNGzaMd955h2effZaePXsSHR3d5Of8uuuu4/PPP+cvf/kLv/zyC+PGjcNms3HgwAE+//xz56zx888/Hx8fH2bMmMFtt91GcXEx7733HtHR0c5KDVetWrWKcePGVRt/J4QQwst5pOe7EEKIM8qMGTMUPz8/paSkpM5jbrjhBsVkMinZ2dnKwIEDlS5dutR7mxMmTFCio6MVi8XiHIlW18eHH35Y621YLBblvffeU2bOnKl07dpV8fX1VQICApShQ4cqL774olJRUVHt+G+++Ubp16+fYjQaa4xH++yzz5ShQ4cqvr6+SkREhHLNNdc4R5BVtWfPHuXSSy9VwsLCFD8/P6VPnz7KE0884by86kg0TWlpqTJ+/HglKChI2bx5c53PiTYSbdu2bfU+d3PnzlUCAwPrvKzqSDRFURSr1aq8+OKLSmJiouLj46N06NBBmTZtWo3RcV9++aVy9tlnK4GBgUpgYKCSmJiozJ8/Xzl48GC961EURXnllVeUoKCgGqPBAGX+/Pn1Xnffvn3K5MmTlaCgICUqKkq55ZZblJ07d1b7GWVnZyvz589XEhMTlcDAQCU0NFQZNWqU8vnnn1e7rfT0dGX69OlKcHCwArg8Hq2u595sNisvvPCC0r9/f8XX11cJDw9Xhg0bpixYsEApKChwHvftt98qgwYNUvz8/JSEhATlhRdeUBYtWqQASlJSkvO4rl27KtOnT691Dfn5+YqPj4/y3//+16U1CyGE8A46RamlZk4IIYQQwo0KCgro3r07//znP7nppps8vZw26bXXXuOf//wnR48erbcZnBBCCO8ie8qFEEII4XGhoaE89NBDvPjiizW6jYuGWSwWXnnlFR5//HEJyIUQoo2RTLkQQgghhBBCCOEhkikXQgghhBBCCCE8xKNB+bp165gxYwZxcXHodDqWLVtW45j9+/dz8cUXExoaSmBgICNGjKgxSkYIIYQQQgghhGiLPBqUl5SUMHjwYN5+++1aLz969Chnn302iYmJrF27ll27dvHEE0/g5+fn5pUKIYQQQgghhBAtz2v2lOt0Or7++mtmzpzp/N5VV12FyWTiww8/9NzChBBCCCGEEEKIVmL09ALqYrfb+eGHH3jooYeYOnUqf/75J926dePRRx+tFrifrqKigoqKimq3k5ubS2RkJDqdzg0rF0IIIYQQQghxJlMUhaKiIuLi4tDr6y9Q99qgPDMzk+LiYv7xj3/w7LPP8sILL7BixQouu+wyfvnlF8aPH1/r9Z5//nkWLFjg5tUKIYQQQgghhBDVnThxgs6dO9d7jNeWr586dYpOnToxZ84cPv74Y+dxF198MYGBgXzyySe13s7pmfKCggK6dOlCUlISwcHBrfoYRE0Wi4VffvmFiRMnYjKZPL0c0QbIa0a4Sl4rorHkNSMaS14zojnk9XNmKyoqolu3buTn5xMaGlrvsV6bKY+KisJoNNKvX79q3+/bty+//fZbndfz9fXF19e3xvcjIiIICQlp8XWK+lksFgICAoiMjJQ/RsIl8poRrpLXimgsec2IxpLXjGgOef2c2bSfuStbqL12TrmPjw8jRozg4MGD1b5/6NAhunbt6qFVCSGEEEIIIYQQLcejmfLi4mKOHDni/DopKYkdO3YQERFBly5dePDBB7nyyis599xzmThxIitWrOC7775j7dq1nlu0EEIIIYQQQgjRQjwalP/+++9MnDjR+fV9990HwNy5c1myZAmXXnop7777Ls8//zx33303ffr04csvv+Tss8/21JKFEEIIIYQQQogW49GgfMKECTTUZ27evHnMmzfPTSsSQgghhBBCiLopioLVasVms9V7nMViwWg0Ul5e3uCxou0xGAwYjcYWGbvttY3ehBBCCCGEEMKbmM1m0tLSKC0tbfBYRVGIjY3lxIkTLRK4Ce8TEBBAx44d8fHxadbtSFAuhBBCCCGEEA2w2+0kJSVhMBiIi4vDx8en3mDbbrdTXFxMUFAQer3X9tcWTaAoCmazmaysLJKSkujVq1ezfsYSlAshhBBCCCFEA8xmM3a7nfj4eAICAho83m63Yzab8fPzk6C8HfL398dkMpGcnOz8OTeVvDqEEEIIIYQQwkUSYAtNS70W5BUlhBBCCCGEEEJ4iATlQgghhBBCCCGEh0hQLoQQQgghhBACnU7HsmXLPL2MRpswYQL33HOPp5fRZBKUCyGEEEIIIUQ7l56ezl133UX37t3x9fUlPj6eGTNmsGbNGk8vzenpp59Gp9Oh0+kwGo0kJCRw7733UlxcXO/1vvrqK5555hk3rbLlSfd1IYQQQgghhGjHjh8/zrhx4wgLC+PFF19k4MCBWCwWVq5cyfz58zlw4ICnl+jUv39/Vq9ejdVqZcOGDcybN4/S0lL+/e9/1zjWbDbj4+NDRESEB1baciRTLoQQQgghhBBNoCgKpWZrnR9lZlu9lzfnQ1EUl9d5xx13oNPp2Lp1K7NmzaJ3797079+f++67j82bN9d5vd27dzNp0iT8/f2JjIzk1ltvrZa1Xrt2LSNHjiQwMJCwsDDGjRtHcnKy8/JvvvmGs846Cz8/P7p3786CBQuwWq31rtVoNBIbG0vnzp258sorueaaa/j2228BNZM+ZMgQ/vvf/9KtWzfnGLLTy9crKip4+OGHiY+Px9fXl549e7Jw4ULn5Xv27GHatGkEBQURExPDddddR3Z2tsvPZ0uTTLkQQgghhBBCNEGZxUa/J1d65L73/W0qAT4Nh3O5ubmsWLGC5557jsDAwBqXh4WF1Xq9kpISpk6dypgxY9i2bRuZmZncfPPN3HnnnSxZsgSr1crMmTO55ZZb+OSTTzCbzWzduhWdTgfA+vXruf7663njjTc455xzOHr0KLfeeisATz31lMuP09/fH7PZ7Pz6yJEjfPnll3z11VcYDIZar3P99dezadMm3njjDQYPHkxSUpIz6M7Pz2fSpEncfPPNvPrqq5SVlfHwww8ze/Zsfv75Z5fX1ZIkKBdCCCGEEEKIdurIkSMoikJiYmKjrvfxxx9TXl7OBx984Azm33rrLWbMmMELL7yAyWSioKCAiy66iB49egDQt29f5/UXLFjAI488wty5cwHo3r07zzzzDA899JDLQfn27dv5+OOPmTRpkvN7ZrOZDz74gA4dOtR6nUOHDvH555+zatUqJk+e7LxvzVtvvcXQoUP5+9//7vzeokWLiI+P59ChQ/Tu3dultbUkCcqFEEK0GXa7wh8pefSODca/9pPjQgghhNv4mwzs+9vUWi+z2+0UFRYRHBKMXt/yu4b9Ta79R9iYMveq9u/fz+DBg6tl18eNG4fdbufgwYOce+653HDDDUydOpUpU6YwefJkZs+eTceOHQHYuXMnGzZs4LnnnnNe32azUV5eTmlpKQEBAbXe7+7duwkKCsJms2E2m5k+fTpvvfWW8/KuXbvWGZAD7NixA4PBwPjx42u9fOfOnfzyyy8EBQXVuOzo0aMSlAshhBD1Wb4njTs//pPwABN3T+pBaNPeZwghhBAtQqfT1VlCbrfbsfoYCPAxtkpQ7qpevXqh0+lapZnb4sWLufvuu1mxYgWfffYZjz/+OKtWrWL06NEUFxezYMECLrvsshrX0/aC16ZPnz58++23GI1G4uLi8PHxqXZ5bSX4Vfn7+9d7eXFxsTPbfzrthIK7SaM3IYQQbcafKfkA5JVaWPD9ARYelP/GhBBCiPpEREQwdepU3n77bUpKSmpcnp+fX+v1+vbty86dO6tdZ8OGDej1evr06eP83tChQ3n00UfZuHEjAwYM4OOPPwbgrLPO4uDBg/Ts2bPGR30nKXx8fOjZsycJCQk1AnJXDBw4ELvdzq+//lrr5WeddRZ79+4lISGhxroaCvhbi7ybEUII0WYcyVQ7vp6XGA3A3jw9ReUWTy5JCCGE8Hpvv/02NpuNkSNH8uWXX3L48GH279/PG2+8wZgxY2q9zjXXXIOfnx9z585lz549/PLLL9x1111cd911xMTEkJSUxKOPPsqmTZtITk7mp59+4vDhw8595U8++SQffPABCxYsYO/evezfv59PP/2Uxx9/vFUfa0JCAnPnzmXevHksW7aMpKQk1q5dy+effw7A/Pnzyc3NZc6cOWzbto2jR4+ycuVKbrzxRmw2W6uurS4SlAshhGgztKD8tvE9CA8wAZCaX+7JJQkhhBBer3v37vzxxx9MnDiR+++/nwEDBjBlyhTWrFnDO++8U+t1AgICWLlyJbm5uYwYMYLLL7+c8847z7m/OyAggAMHDjhHrN16663Mnz+f2267DYCpU6fy/fff89NPPzFixAhGjx7Nq6++SteuXVv98b7zzjtcfvnl3HHHHSQmJnLLLbc4M/5xcXFs2LABm83G+eefz8CBA7nnnnsICwvz2DYDndLUnf9tRGFhIaGhoRQUFBASEuLp5ZxxLBYLy5cv58ILL8RkMnl6OaINkNeMqEup2eocO/PHE1O4fuEW9pwq5N2rh3DBoE4eXp1oC+Tvi2gsec2IqsrLy0lKSqo2H7s+drudwsJCQkJCPLqnXLSe+l4TjYlD5dUhhBCiTTiWpZ7hjgj0ISLQh05h6n9+J/PLPLksIYQQQohmkaBcCCFEm3A4swiAnh3UESadw9XuqlK+LoQQQoi2TIJyIYQQbYK2n7xnjBqUdwpTg/KTeZIpF0IIIUTbJUG5EEIIr2W22ikzq51QnUG5I1PeyZkpl6BcCCGEEG1X7ZPuhRBCCA+z2uxMfW0diqKw/K/nVAbl0Y7ydceecgnKhRBCCNGWSVAuhBDCK53MKyMpW23u9tm2ExzPKQUqg/I4R/l6QZmVwnILIX7SGVkIIYQQbY+UrwshhPBKybmlzs9fXXUIm10h0MdAx1A1Qx7kayTQqE71TJV95UIIIYRooyQoF0II4ZVSckqcnxeWWwE1S67T6Zzfj/BV/5Vmb0IIIYRoqyQoF0II4ZWSHeXqPsbK/6p6OErXNRG+aqb8ZF4pQgghhBBtkQTlQgghvJJWvn7juAQMejU73rNGUK7+K5lyIYQQwjssWbKEsLAwTy+jTZGgXAghhFdKcWTKx3SPZM7IePQ6OLdXh2rHSKZcCCGEaNjTTz+NTqer9pGYmFjtmPLycubPn09kZCRBQUHMmjWLjIwMD624fsePH0en02EwGEhNTa12WVpaGkajEZ1Ox/Hjx2tcd+rUqRgMBrZt2+am1TZMgnIhhBBeR1EUUhyZ8q6Rgfzt4gHsenoqAzqFVjsuQu35JplyIYQQogH9+/cnLS3N+fHbb79Vu/zee+/lu+++44svvuDXX3/l1KlTXHbZZR5arWs6derEBx98UO1777//Pp06dar1+JSUFDZu3Midd97JokWL3LFEl0hQLoQQwutkFVdQZrGh10GnMH/0eh1BvjWneFZmyiUo1yiKwsYj2TyxbA87TuR7ejlCCNG+KQqYS+r+sJTWf3lzPhSlUUs1Go3ExsY6P6KiopyXFRQUsHDhQl555RUmTZrEsGHDWLx4MRs3bmTz5s313u6SJUvo0qULAQEBXHrppeTk5FS7/OjRo1xyySXExMQQFBTEiBEjWL16tfPyv/3tbwwYMKDG7Q4ZMoQnnnii3vueO3cuixcvrva9xYsXM3fu3FqPX7x4MRdddBG33347n3zyCWVl3vH+QeaUCyGE8Dpa6XrHUP9qjd5Op+0pLyiznPGzyksqrHz1x0ne35TMkcxiADKLyvn3dcM9vDIhhGjHLKXw97haL9IDYa1534+dAp9Alw8/fPgwcXFx+Pn5MWbMGJ5//nm6dOkCwPbt27FYLEyePNl5fGJiIl26dGHTpk2MHj261tvcsmULN910E88//zwzZ85kxYoVPPXUU9WOKS4u5sILL+S5557D19eXDz74gBkzZnDw4EG6dOnCvHnzWLBgAdu2bWPEiBEA/Pnnn+zatYuvvvqq3sd08cUX8+677/Lbb79x9tln89tvv5GXl8eMGTN45plnqh2rKAqLFy/m7bffJjExkZ49e/K///2P6667zuXnsLVIplwIIYTX0Tqvd40MqPc4PwOEB6iB+Jk6q/xYVjFPf7uX0X9fwxPf7HUG5ADFFVYPrkwIIYS3GDVqFEuWLGHFihW88847JCUlcc4551BUVARAeno6Pj4+NRq0xcTEkJ6eXuftvv7661xwwQU89NBD9O7dm7vvvpupU6dWO2bw4MHcdtttDBgwgF69evHMM8/Qo0cPvv32WwA6d+7M1KlTq2W8Fy9ezPjx4+nevXu9j8tkMnHttdc6S9EXLVrEtddei8lU8yT96tWrKS0tda7v2muvZeHChfXevrtIplwIIdzEbLWz4Ug2I7pF1FqKLSol57oWlINa3p5XauFkXhl9O4a09tK8gt2usPZQJks2JrPuUJbz+92iArl+TFeCfI08+L9dlFvsHlylEEKcAUwBasa6Fna7ncKiIkKCg9HrWyEXamr4/0jNtGnTnJ8PGjSIUaNG0bVrVz7//HNuuukml26jf//+JCcnA3DOOefw448/sn//fi699NJqx40ZM4YVK1Y4vy4uLubpp5/mhx9+IC0tDavVSllZGSkpKc5jbrnlFubNm8crr7yCXq/n448/5tVXX3VpXfPmzWPs2LH8/e9/54svvmDTpk1YrTVPSi9atIgrr7wSo1F9DzZnzhwefPBBjh49So8ePVy6r9Yi7wqFEMJNPv/9BI8v28Nt53bn0Qv7eno5Xi0lpwSALhENl+V1CvNjz6lCUs+QDuybjubw8Je7nI3wdDqY1Cea68cmcE7PKPR6HWsPZgJQbrF5cqlCCNH+6XR1l5Db7WCyqZe3RlDeDGFhYfTu3ZsjR44AEBsbi9lsJj8/v1q2PCMjg9jYWACWL1+OxWIBwN/f3+X7euCBB1i1ahUvvfQSPXv2xN/fn8svvxyz2ew8ZsaMGfj6+vL111/j4+ODxWLh8ssvd+n2Bw4cSGJiInPmzKFv374MGDCAHTt2VDsmNzeXr7/+GovFwjvvvOP8vs1mY9GiRTz33HMuP57WIEG5EEK4SWZhOQC7ThZ4eCXeT8uUd4loOAvQOVx9Y3AmNHvLKCzn9qXbyS+1EOJn5MoR8Vw7uitdI6u/IfQzGQAJyoUQQtSuuLiYo0ePOvdTDxs2DJPJxJo1a5g1axYABw8eJCUlhTFjxgDQtWvXGrfTt29ftmzZUu17pzeG27BhAzfccIMzo15cXFxjVJnRaHQ2bfPx8eGqq65qVOA/b9487rjjjmoBd1VLly6lc+fOLFu2rNr3f/rpJ15++WX+9re/YTAYXL6/liZBuRBCuInZpnZJPZpV3MCR4kQjy9eh/QfldrvCA1/sJL/UQv+4ED6/bQyBdWyDqAzKpXxdCCGEmq2eMWMGXbt25dSpUzz11FMYDAbmzJkDQGhoKDfddBP33XcfERERhISEcNdddzFmzJg6m7wB3H333YwbN46XXnqJSy65hJUrV1YrXQfo1asXX331FTNmzECn0/HEE09gt9f8/+nmm2+mb1+1knDDhg2Neny33HILV1xxRY098ZqFCxdy+eWX1+jyHh8fz6OPPsqKFSuYPn16o+6zJXlXHYUQQrRjZqv6H1BmUQVF5RYPr8Z7FVdYyS5WS9q6uBKUa5ny/PZdvr5oQxLrD2fjZ9Lz+lVD6wzIAfxM6n/vFVbJlAshhICTJ08yZ84c+vTpw+zZs4mMjGTz5s106NDBecyrr77KRRddxKxZszj33HOJjY1tsPv56NGjee+993j99dcZPHgwP/30E48//ni1Y1555RXCw8MZO3YsM2bMYOrUqZx11lk1bqtXr16MHTuWxMRERo0a1ajHZzQaiYqKcu4Xr2r79u3s3LnTWQFQVWhoKOedd57HG75JplwIIdzEYqs8K5yUXcKgzmGeW4wX08ahhQeYXBpx1jnMD2jfmXJFUXhjzWEA/m96P3pGB9V7vJ9RMuVCCCEqffrppw0e4+fnx9tvv83bb7/dqNueN28e8+bNq/a9+++/3/l5QkICP//8c7XL58+fX+N2FEXh1KlT3HHHHQ3eZ0JCAko9c9qHDBnivLyhY5cvX97g/bU2yZQLIYSbVA3KpYS9brtO5gPQJdK12atxjvL1/FJLu61AKCy3UliudpK9YljnBo+XPeVCCCHakqysLN566y3S09O58cYbPb0ct5NMuRBCuIlWvg5wLKvEgyvxXukF5Tz/4wEAzkuMduk6Qb5GwgNM5JVaSM0vIzG24ex6W5NVVAFAsK/RGXDXRytft9oVrDY7RoOcgxdCCOG9oqOjiYqK4j//+Q/h4eGeXo7bSVAuhBBuYrZJUJ5ZWM7KvenMHhGPr7F6cGm3K9z/xQ4KyiwM7BTKX8a7PjO0c3gAeaUFnMwtIzG2/c0qzy5Wg/IOwb4uHV81cC+32gmSoFwIIYQXq6+8/Ewg/0sLIYSbSPk6vPnzEZ74Zi/vrD1a47JFG5LYcCQHf5OB164ago/R9f+iKseitc9mb1qmPMrFoNy3ynMnJexCCCGEd5OgXAgh3KRq+XpSdgk2+5l3VlgLmpfvTqv2/X2nCvnnioMAPH5RX3p0qL+R2ena+6xyLSh3NVOu0+mcgbkE5UIIIYR3k6BcCCHcxGKrDMIrrHZO5bfPALI+uaVqI7ZDGcXOaoFyi42/fvonZpudyX1juHpkl0bfbudwdXRauw3KtfL1INeCcpBZ5d6kzGzj2BlaHSOEEKJhEpQLIYSbVN1TDmdmCXtuSYXz8xV70gH4x48HOJxZTIdgX16YNRCdTtfo2+3czmeVNzZTDpXN3iRT7nn3fraDSS//yp7UAk8vRQghhBeSoFwIIdxEK1/X9kqfic3e8koqR5at3JvOLwczWbLxOAAvXj6IyEZkgqtq95nyJgXlaqa8wipBuaf9eSLP8W++ZxcihBDCK0lQLoQQbqI1eusdo+6XPtMy5RVWG8UVVufXu04WcN9nOwC4YWwCE/q4NgKtNp3C2/es8uymlK8bpXzdG5RbbGQUqj+/5Owz70ScEEKIhklQLoQQbqIF5X1i1JFdZ1qmPN+xn9yg1zEyIQKAvFILvWOCeGRaYrNuW5tVDpDaDvfqS/l625WSW7ml4njOmfU7L4QQdbnhhhuYOXOmp5fhNSQoF0IIN9HK1/t2DAZg76mCapnj9i63xAxAeIAP0wbGAuBj0PP6VUOrzdVuKmcJe277CsptdoUcx3PXmKDcVxq9eYXknKpBefvseSCE8H7PP/88I0aMIDg4mOjoaGbOnMnBgwerHVNeXs78+fOJjIwkKCiIWbNmkZGRUe2YlJQUpk+fTkBAANHR0Tz44INYrd75XmbJkiXodDr69u1b47IvvvgCnU5HQkJCjcvKysqIiIggKiqKioqKGpe3BgnKhRDCTbTu60O7hJMQGUBhuZV//1pzXnd7lecILCMCTVwxPJ5ZZ3Xm9auG0LdjSIvcfnudVZ5XasZmV9DpICLQx+XrVXZfl0y5JyVXyY6n5JSekaMQhRCe9+uvvzJ//nw2b97MqlWrsFgsnH/++ZSUVP6Nuvfee/nuu+/44osv+PXXXzl16hSXXXaZ83Kbzcb06dMxm81s3LiR999/nyVLlvDkk0964iG5JDAwkMzMTDZt2lTt+wsXLqRLl9qnvXz55Zf079+fxMREli1b5oZVSlAuhBBuo3VfD/Ax8Mg09azte+uPnTGj0XJLKzPlQb5GXp49mGkDO7bY7bfXWeVa6XpEgA8mg+v/bTvnlEujN486UaV83Wyzk15Y7sHVCCFamqIolFpK6/wos5bVe3lzPhTF9ZN8K1as4IYbbqB///4MHjyYJUuWkJKSwvbt2wEoKChg4cKFvPLKK0yaNIlhw4axePFiNm7cyObNmwH46aef2LdvHx999BFDhgxh2rRpPPPMM7z99tuYzeY679tms3HfffcRFhZGZGQkDz30UI21r1ixgrPPPtt5zEUXXcTRo5WJi0mTJnHnnXdWu05WVhY+Pj6sWbOmzvs2Go1cffXVLFq0yPm9kydPsnbtWq6++upar7Nw4UKuvfZarr32WhYuXFjnbbcko1vuRQghhLN83WTQM7V/DCMTIth6PJeXVh7klSuHeHZxbpDrzJS7nu1tjPbagb0p+8lB5pR7i+Tc6pUbydkldArz99BqhBAtrcxaxqiPR3nkvrdcvYUAU0CTrltQoI5ojIhQe7xs374di8XC5MmTncckJibSpUsXNm3axOjRo9m0aRMDBw4kJibGeczUqVO5/fbb2bt3L0OHDq31vl5++WWWLFnCokWL6Nu3Ly+//DJff/01kyZNch5TUlLCfffdx6BBgyguLubJJ5/k0ksvZceOHej1em6++WbuvPNOXn75ZXx91f8PP/roIzp16lTtdmozb948JkyYwOuvv05AQABLlizhggsuqPY4NEePHmXTpk189dVXKIrCvffeS3JyMl27dnXxmW0ayZQLIYSbaI3efAx6dDodj1+kZsu/+jOV3Sfb//xi557yVgvK2+escmfn9cYG5UZp9OYNUhz7yIN81TxIkjR7E0J4mN1u55577mHcuHEMGDAAgPT0dHx8fAgLC6t2bExMDOnp6c5jTg9kta+1Y2rz2muv8eijj3LZZZfRt29f3n33XUJDQ6sdM2vWLC677DJ69uzJkCFDWLRoEbt372bfvn0AzjL6b775xnmdJUuWcMMNN6DT6ep9vEOHDqV79+7873//Q1EUlixZwrx582o9dtGiRUybNo3w8HAiIiKYOnUqixcvrvf2W4JHM+Xr1q3jxRdfZPv27aSlpfH1119X68J3ww038P7771e7ztSpU1mxYoWbVyqEEM3nDModwdKgzmHMHBLHsh2nePaHfXx66+gG/2Npy5x7ygMkU94Yzkx5I2e4O+eUS1DuMTa7wglHj4OxPSL5aV9GtcZvQoi2z9/oz5art9R6md1up6ioiODgYPT6ls+F+hubVnUzf/589uzZw2+//dai60lJSaFfv37Orx977DHmz59PWloao0ZVVhMYjUaGDx9erYT98OHDPPnkk2zZsoXs7GzsdrvzNgcMGICfnx/XXXcdixYtYvbs2fzxxx/s2bOHb7/91qW1zZs3j8WLF9OlSxdKSkq48MILeeutt6odY7PZeP/993n99ded37v22mt54IEHePLJJ1vlZ6jxaFBeUlLC4MGDmTdvXrUmAlVdcMEF1c5OaOUKQgjRltjtirPRm8lQGXg/eEEiP+5JZ0tSLqv2ZXB+/1hPLbHV5TpGorVW+frps8qD/Uytcj/upgXlUY0uX9f2lEv5uqekFZRhsSmYDDrGOILy4zKrXIh2RafT1VlCbrfbsRqtBJgCWjWga4w777yT77//nnXr1tG5c2fn92NjYzGbzeTn51fLlmdkZBAbG+s8ZuvWrdVuT+vOHhsbS1xcHDt27HBeppXGu2LGjBl07dqV9957j7i4OOx2OwMGDKi2V/3mm29myJAhnDx5ksWLFzNp0iSXy8qvueYaHnroIZ5++mmuu+46jMaaYfDKlStJTU3lyiuvrPZ9m83GmjVrmDJlisuPp7E8+uqYNm0azz77LJdeemmdx/j6+hIbG+v8CA8Pd+MKhRCiZVjslYGRyVj5p7dTmD83nd0NgH/8eMCZTW+P8lp5T3l7nVWeVdy8TLmUr3uOVroeHx5A9w5BAJIpF0J4hKIo3HnnnXz99df8/PPPdOvWrdrlw4YNw2QyVWuadvDgQVJSUhgzZgwAY8aMYffu3WRmZjqPWbVqFSEhIfTr1w+j0UjPnj2dHxEREYSGhtKxY0e2bKmsJrBarc4GcwA5OTkcPHiQxx9/nPPOO4++ffuSl5dX4zEMHDiQ4cOH89577/Hxxx/XWYJem4iICC6++GJ+/fXXOq+3cOFCrrrqKnbs2FHt46qrrmr1hm9e3+ht7dq1REdHEx4ezqRJk3j22WeJjIys8/iKiopq8+QKCwsBsFgsWCyWVl+vqE57zuW5F65qr6+Z0irzyPV2G1Uf3s3juvLZthMcyy7hg41JXD+69hEdbV2OI7gM8dW3yM+3ttdKpzB/8kotHM8qokdk+2imleno1h0eYGzU8+ZIlFNaYW13v09N5e6/L8eyigCID/enU6h6Mup4TgkVFWb0+va7VaU9aa//J4mmsVgsKIqC3W53llfXRyvP1q7jSfPnz+eTTz7h66+/JjAwkFOnTgEQGhqKv78/wcHBzJs3z9klPSQkhL/+9a+MGTOGkSNHYrfbmTx5Mv369ePaa6/lhRdeID09nccff5w77rgDk8lU52O8++67+cc//kGPHj1ITEzk1VdfJT8/3/m8hIaGEhkZyb///W9iYmJISUnhscceA6jxXM+bN4+7776bwMBALrnkknqfV+0y7d9Fixbx1ltvERkZid1ud/587HY7WVlZfPfddyxbtqxaCT6oJeyzZs0iOzu7RvZfux2LxYLBYKh2WWP+bnh1UH7BBRdw2WWX0a1bN44ePcpjjz3GtGnT2LRpU40HrXn++edZsGBBje//9NNPBAQ0rTuhaL5Vq1Z5egmijWlvr5liC2h/clf9tJLT349PitbxRZKBl1fuJyBzDwFe/de5adJyDYCOfX9upehwy91u1deKoVwP6Fm1YTsVx9rHPOikNPV5O7ZvB8tT/3T5esdO6QADx5JPsHx5cqutry1y19+Xtcnq69FelMmuTenodQYqrHY+/eZHwmQ3XpvS3v5PEk1jNBqJjY2luLi43hFgpysqKmrFVbnm3XffBajRqfztt992jgZ7+umnsVqtXH755ZjNZiZNmsRLL73kTHICLF26lPvvv59x48YREBDAnDlzuP/++6sdc7qbb76Z5ORkbrjhBvR6Pddeey3Tp0+nsLDQeb3//ve/PPLIIwwaNIiePXvywgsvcNFFF1FWVlbttqdPn859993HZZddhtlsrvfnUF5ejqIo1a5vMpmcX5eXl2O32yksLOS9994jICCAESNG1HgsI0aMwM/Pj4ULF3LbbbdVu8xsNlNWVsa6deuwWq3VListdb0ySqc0ZsBdK9LpdDUavZ3u2LFj9OjRg9WrV3PeeefVekxtmfL4+Hiys7MJCQlp6WWLBlgsFlatWsWUKVMwmdrH/k7RutrrayajsJyzX1yHQa/jwIKae5KsNjsz3t7EkawSbhrXlUcu6OOBVbYeRVEY8Lc1mK121t5/TouMhKrttfKPFQdZuCGZeWO78ui09vEcjnz+F/JKLSy/cyy9YoJcvt7HW0/w1Hf7mdI3mn9dPaT1FtiGuPvvy92f7uTHvRk8Nq0PN47tyuRXfyM5t5SP5g1nVDfX91oKz2mv/yeJpikvL+fEiRMkJCTg5+fX4PGKojgbvbXnRq7udPz4cXr16sWWLVs466yzPL0cysvLOX78OPHx8TVeE4WFhURFRVFQUNBgHNqmcjHdu3cnKiqKI0eO1BmU+/r61toMzmQyyR9TD5LnXzRWe3vNKDq1hMnHoK/1cZlM8H8X9ePGxdv4cPMJ5o7tTpfIlq/u+WZHKj4GPdMGdmzx265PSYXVOac9OjQAk6nl/vup+lrpEqkGracKKtrF68dstZPnaJDXMTywUY8p0E8tl66wKe3iuWhJ7vr7csLR26Bbh2BMJhPdOgSSnFvKyfwKzpafSZvS3v5PEk1js9nQ6XTo9XqXGrdpZdPadUTTWSwWcnJyePLJJxk9ejTDhw/39JIA0OvVMbe1/Y1ozN+MNvXqOHnyJDk5OXTs6N43k0II0VxmRwO3qp3XTzehdwfO6RWF2WbnpZ8OtvgalmxI4q+f7uD2pX9wIL3uMrPCcgtJLdwhWptR7mvU42+qfftRS2hvs8pzStTKL5NBR6h/4wICZ/d1afTmEYqiOJu6dXWcYEuIDARkVrkQQjTWhg0b6NixI9u2bXOW4rcnHs2UFxcXc+TIEefXSUlJ7Nixg4iICCIiIliwYAGzZs0iNjaWo0eP8tBDD9GzZ0+mTp3qwVULIUTjnT6jvDY6nY67JvVi/eFsNh3LadH7/3F3Ggu+3+f8+vXVh3nn2mHVjskoLGfhb0ks3ZxMidnG93edzYBOoS1y/3mlalAeGejTqiV87WVW+ZHMYj7anEyGo8lbVJBvoxuD+RllTrkn5ZVaKCpX9xfGO16XPaLVSo59p+o+KSaEEKKmCRMm4CW7rluFR4Py33//nYkTJzq/vu+++wCYO3cu77zzDrt27eL9998nPz+fuLg4zj//fJ555hmZVS6EaHMsVvU/Eh9D/QVKXSLUN++5JWbsdqVZHZrLLTa+3XmKXw9lsWpvBooCU/rFsHp/Bj/uSWffqUL6xYWQlF3Cf9Yd5cvtqc6MPsC+tMIaQXlRuYX0gnJ6xQQ3ai1apjy8lcahaVyZVX4wvYgOwb6tNpqtuf5MyWPuoq0Ullc2jImPaPxWhsqRaO13zJ430wLvrpEB+PuoP4vhXdWxrn8k52G12TE28PdACCHEmcGjQXlDZzxWrlzpxtUIIUTrMdvUbKWpnkw5VM7wttkVCsoszQpi//b9Pj7ekuL8+oL+sbx9zVn89dM/+X5XGs98v4+IIB9+3J2G3fGneERCOIoCvyfnkV5QXu32MgvLufRfGzlVUMbq+8bTo4PrTce0THlrB8LarPK8Ugup+WUkxlYG5ceyivn78gOs3p/BwE6hfHfX2a26lqbYeDSbm9//nVKzjcHxYUxOjEav13HBgNhG35azfN0qmXJP2JWaD1DtxFafmGCC/YwUlVvZn1bEwM4tU4kihHCv9pyxFY3TUq+FNtXoTQgh2iqzI1NuaiAz5mPUE+JnpLDcSk5JRbOC8hTHftaLB8cxd2xXzuoSjk6n457Jvfhhd1q1EvlJidHcPqEHIxIieG31IX5PziOtoLIEvLjCyg2Lt5HqaFy1+2RBo4Ly3BK1WVl4QOtnpzuHB5BXWsDJ3DISY9Vup1uTcrn6vc1YHWcf9p4qoNxic2aTvcHqfRnc8fEfmK12zu4Zxb+vG0agb9P/m67MlEtQ7gl7UgsAGFglKNfrdQzvGs4vB7PYejxXgnIh2hitcVdpaSn+/s2fIiLaPm3sWXMbQUpQLoQQbuDcU+5CuWpUkC+F5Vayi830jG76fRZVqOXPFw+OY1jXyvFLPaODmTsmgQ82HWfG4Dj+Mr4HfTtWjuroGKqO9EhzZMoVRWH+0j/Yl1a5D1ZrYOWqXEfDMneUjHcO92d3agEn8yrX+NUfJ7HaFYZ1DedQehFFFVaSskuqPW5P+mZHKvd9vhObXWFKvxjenDO02ScMKhu9Sfm6J+x2BOWDTtsCMqJbBL8czOL347ncdHY3TyxNCNFEBoOBsLAwMjMzAQgICKi3T4rdbsdsNlNeXi7d19sZRVEoLS0lMzOTsLAwDIbm/Z8tQbkQQriBNg6sofJ1gMggH45ll5BTbG7WfRaXq9npIL+af+qfmtGPx6f3rXVPa8dQ9ey/Vr5+IL2IXw9l4WPUc15iND/uSSc5t3Hdo92bKXd0YK/S7O1gRhEAN4xNYMnG42xPzuNIZrFXBOVLtyTz+LI9KApcOrQT/7x8UIMVFa7wNUqm3FPySsycyFVff/1PD8oT1BNk247noiiKzC4Woo2JjVW3E2mBeX0URaGsrAx/f3/5XW+nwsLCnK+J5pCgXAgh3KAyU97wf8qRgWozS20cVlNpnZ+DaimB1ul0GOtYi5YpP+UoVdfGow2IC2HawI78uCfdWRrvqrwSbU9568/5Pb0Du6IoHMkoBqB3TDA9OwQ5g3JPe/fXo/zjxwMAXDu6C3+7eECzmvtVpWXaK6x2Cf7cbM8pNUveNTKgxii7QZ1D8THqyS42czynlG5RgZ5YohCiiXQ6HR07diQ6OhqLxVLvsRaLhXXr1nHuuefKnPt2yGQyNTtDrpGgXAgh3MDswkg0TVSwmk3Obm6m3FG+HlxLprw+sY6gvLDcSkmFleOOmcoJkYF0dXQBT8ltZPm6s9Fb60/POH1WeVpBOUUVVox6Hd2iAunpGEt1JMuzQfm/1h7hnyvUefS3T+jBQ1P7tGjgrJWvgxqYe9P++fZudy37yTW+RgODO4ey7Xge25JyJSgXoo0yGAwNBmQGgwGr1Yqfn58E5aJesrlBCCHcwFm+7kJZsjNTXtz0TLnNrlBqVsuWaxsLVp9gPxPBjux6emE5xx2Z8q6RgXSNVIPyzKIKysyul0XnOUeiuS9Tnpxdit2ucMhRup4QFYiPUV8ZlGd4Lij/cvtJZ0D+4NQ+PHxBYotnsqsG4VLC7l61NXmraniVEnYhhBBCgnIhhHADi8217usAUUFqprw5e8q1LDlAoG/jM6Ratjy9oJzjjlL1hKgAwgJ8CHFk3l3NliuKQnax+xq9de8QSKCPgaIKK/vTCznsLF1Xg3EtKE/KLsFqc38TtN8OZ/Pwl7sAuPXc7syf2LNV7sdk0GNwlMJLszf32nWy/qB8pCMo33g0xyOvQSGEEN5FgnIhhHADSyPK1yODmr+nXAvKfYx6Z8Ovxoitsq88uUr5OqgZc8D5/YZsO55HXqmFAB8DXRzl763JZNAzopsa9Gw6muPMlPeOCQagU5g/fiY9ZpudE1WawbnLU9/uwWpXmDE4jkcuSGzV+/Izah3YJVPuLnklZmc/g9ObvGlGdIsgLMBEan4ZH21OdufyhBBCeCEJyoUQwg208nVXRqJFBjZ/T3mxo8lbcBPnXMc5OrAfyy4ho1A9OaAF5V0iG7ev/NOtKQDMGBRHgI97WpmM6R4JwOZjuRzKrGzyBuqsaG3GurubvZWarRzNUk9mPD2jX4s1dauLs9lbWTEc/BF+exX++BAO/QRpO6EoHeySqW1JWpO3hFqavGmCfI08cH4fAF5ZdahZW1WEEEK0fdLoTQgh3EBr9GZypfu6I1Oe3Yw36kX1jENzhZYp33IsB4CwABOhAWqAoWW7XZlVXlBq4YfdaQBcNTK+SWtpijE91KB8S1IOdru6dUArXwe1hH3vqUKOZBYzpV+M29alnQSICvJx/pxbjaIwTr+XqaYf6LVkF9jKaz/OPwJ6TIQe56n/hsS17rraOa10fUAdWXLNnJFd+GRrCntPFfLiyoP8Y9YgdyxPCCGEF5KgXAgh3KAx5evanvKicisVVluTys+LKuoeh+YKbSyaFmBoJeuAswN7sguZ8mU7Uqmw2kmMDWZIfFiT1tIU/TqGEOxrdI6FMxl01R5DT0em/HBmkdvWBHAwXb2/XtHBLX/jigJ5SXBiG5zYAkm/8oblCBgAGxAaD/GjoLwAitOhKANKsqAsF/Z8qX4AdOgLPR0BetezweTX8mttx7Qmb4M61x+UG/Q6Flzcn8vf3cRnv5/g2tFdGwzkhRBCtE8SlAshhBs0pvt6iJ8Jo16H1a6QW2Kmo6OUvDGc5etNzJR3DFPv0+rIMneLrNwL7ixfb2BPuaIofOIoXb9yRLxb52QbDXpGdotgzYFMALpHBVV77rVmb0fdXL5+OLN607lmsZTBqT/VAPzENji5VQ2yqyjFn/9Zz6bf9PkMHz0BTv8Z2Cxw8nc4+jMcXQOpf0DWfvVj01tqIH/xm2qALlyijUNzJcAenhDBjMFxfLfzFK+tPsx/5w5v7eUJIYTwQhKUCyGEGzgz5S4E5Xq9jsggHzIKK8gpbmJQ7syUN20EmZYp11TLlDs+P5lXhs2uODt8ayw2Oz/tzeD9jcc5kF6Ej1HPpUM7NWkdzTGmR6QzKO91WhCsfX00qwRFUdx2wkBrOtcrphmZ8uMbYNUT6p5wu7X6ZQYf6DgE4kdC5xHc8ksgG06aeTe4b82AHMBggq5j1I9J/weluXBsrRqkH1oJBSfgw5kw7Ea44B+SNW9A1SZvrma975ncix92nWL1/gx2nyxgYAMZdiGEEO2PBOVCCOEG2kg0V8rXQZ1VnlFY0eR95c3NlMeeFpQnRFVmymND/PAxqN3LT+WXEe8oZ88qquDTrSks3ZJCeqG6f9mo1/HA+b0JC2j9UWinG+1o9gbQ57QguGtkIEa9juIKK09/u5fz+sYwsltEtdneraFyPFsTg/L8E/Dp1VCer34dFKsG4PEj1dL0joPBWLlXXdm4Gcihwupi9/WACBhwmfpRUQyrn4Jt/4XtiyFzP1z1MQRGNnw7ZygtS94tKpAQP9dOiPXoEMTMIZ346s9UXl19iEU3jGjNJQohhPBCEpQLIYQbVDSifB0gspmzyp2N3pq4pzzY10igj4ESsxrMVc2UG/Q6Okf4cyyrhJTcUrKLK/hgUzI/7EpzNrSLCvLh6pFduHpU1xoBvrv07RhCiJ+RwnJrjcy0yaBnWNdwtiTl8v6mZN7flIy/ycDYHpFM6NOBCX2inScbWkpxhZXUfDWL2qTydZsVvrxZDcg7DYMrlqjl5fVk+bWTDE0aieYbBNNfhj4Xwhc3wonNsHAyXPZf6Dys8bd3BmhM6XpVd53Xi292nuLnA5nsOJHv1v4LQgghPE+CciGEcAOLrXFBeVQzZ5U7G701MVOu0+noGObv7BaeUCUoB7UD+7GsEv766Z/VRrcN7RLG3DEJTBsY26QGdS3JoNfx2IV9WX84mwl9OtS4fMmNI1l7MJO1B7NYeyiTjMIK1hzIdJS87+Xc3h14/8YRLVbafthRut4h2LdplQO/vqAGxr4hMGshhHVp8Cp+JvX1pp0UapKe58FNP8HSKyD3GPx3EvSeBuc9ATH9m367Xi6zsJzrFm7lyhHxzDu7m0vX0Zq8DewU0qj76hYVyKVDO/G/7Sd5ddUh3p83stHrFUII0XZJUC6EEG7QmO7rUDmrvKmZ8uaWr4O6r/xIZjEhfkbCA6qX4qpBehbZxWZ8DHpmDI5j7tiuDOoc1uT7aw1XjezCVSNrD179fQxMG9iRaQM7oigK+9OKWHtIDdK3JuWy7lAWeaUWIgJbpvRe209+eim9S/Z9C+v+qX5+0asQ4VqQ6GdsRqa8quhEuGUNrHoKdn0Kh36Ewz/B6NthwiPg2wrd5D1s7aEsDmYU8b/tJ10OyrVpBQM7hTX6/u6a1JOv/0zl10NZbE/OY1jX8EbfhhBCiLbJtXeHQgghmkXrvu7jwpxyqDqrvIlBuSNTHtzE8nVQ944DJEQF1sgWXz+mK5P7RvPg1D5senQSL88e7HUBeWPodDr6xYVwx4SefH7bGGeGuaTC2sA1XXfIsZ/89KZzDTq1A76+Tf185G0w8HKXr+rrLF9vRqZcExQNl74D87dC3xmg2NQO7a8OgM/nwu+L1Ey6ojT/vrzACcfIP1f7OuSVmJ3bE/o3MlMO6haRy8/qDMBrqw81+vpCCCHaLsmUCyHOKO7stF1VY8vXnXvKm1i+XtzM8nWAOMdYtK6nla4DdO8QxH/ntt+GVEG+RsotZufz2BK0THmjmrzlJcMnc8BSCj3Og6l/b9R9+joqM5qdKa8qqhdc+REcXg3LH1Bno+9bpn4AhHaB7uOh+wToMUltHtcGpTiC8twSM3a7gl5f/9+NpjR5O92dk3ry5R8nWX84m23HcxmR0DafOyGEEI0jmXIhzhDvrD3K+Bd/ISWn1NNLaVUZheX8kZJX62X3f76Tya/82rIBiovMjey+HtXsRm/NG4kGMOuszlw4MJabXCzdbU8CHRUGLZkpr+y87mKmPOcoLL4Qik5BVB+4YjEYGneSxa8lM+Wn6zUZ7vwd5v0EEx6DruNAb4KCFPjzQ/jyJnhtEBxe1fL37QZaUG61KxSUWRo8frdzP3nTR5rFRwRwxfB4AF5dJdlyIYQ4U0imXIgzxPe7TpGcU8pHW5J57MK+nl5Oq7n1w+3sPJHP6vvOpWd0ZUbSarPzzY5UrHaFY1kl9ItrfHlpc5gdI6ka2+itqSPRmtt9HaBLZAD/uubM7LId6KM+by2VKS8oszjHxDlfl3Y75CdDWS6UF9T8+HMpFKdDVG+4/hvwa3ywp5Xhl7s6Eq2xDEboMkr9mPAwmEsgeRMc+0Xdc559CD6+Ei78JwybB/q2kwvQytdB/T0Mb6C3wO6TzQ/KQc2W/2/7CTYezWHzsZxqo/2EEEK0TxKUC3GGyC9Vg7Rvd5zi4QsSMTRQitkWKYrCoXS1RPiP5PxqQfnJvDKsdjVbrY3tcidtTrnr5euO7uvF5lpL7m/78HeOZ5fy5R1jaw28nXvKm1G+fiYLcmbKWyCYtds4kV2IARujA9MI3fEfOL4BUjZCWe1VHU7R/dWAPKhm93hXNGskWlP4BKoZ9F6T4byn4Lu/ws6P4Yf7YfUCdY56/5kw5Bow+btnTU1QUmGt1s8hu9hMr5j6r9PUcWin6xTmz5Uj4vlocwqvrjrEZ7eNadbtCSGE8H7ybk2IM0R+qfoGM72wnK1JuYzp0f6yL4VlVsocwccBR3CuScoucX5u8UhQrt6nbyO7r5ttdooqrNX2qBaVW1i5NwOAFXvSuXxY5xrXb4nu62cybS9+cUXDZcvVlORA+i71I83xb84RBih2jvoBNmBlleMNvhDYQc2Cn/4R0hGG3QD+Te/C7ed4vVW0Rvl6Q4w+MPNfENUT1r0EFYVwfL36sfYFGDMfhs8Dv9qrVjzV/wHgRF71bT4NVazkVmnyNqAJTd5ON39iTz7fdpItSblsPJrN2B5Rzb5NIYQQ3kverQlxBjBb7ZSYKzNl3+xIbZdBeVphmfPzA+mF1S6rGpSbmzOzuYm0+3Q1U+5nMhDka6S4wkp2UUW1oPxoVuVj+WZHao2g3GZXnD/v5pSvn8m0PeXF9WXKK4rg2K/VA/DC1Hpvt1TnT0CPsyFhHHQ9G+KGgKHp+/4b4vZM+el0Ojjnfhj7V8jarz5fW96FghOw+in47RU463q1RD8gEgKiHP9G8PK6DJZsSuG7u86mW1TNZoOt6fTeGw0F5VqWvHtUIMFNbPJWVcdQf+aMjOf9Tcm8uuoQY7pHeuwEhRBCiNYn79aEOAOc3qRo+e40FlzSH1/HDOP2Iq2g3Pn5wXoy5Z4oXzc7u6+7/sY6OsSX4iwrGYUVdO9Q2RzsSGax8/MNR7LJLConOtjP+b0Sc+U+6OZ0Xz+TBfmqvxu1Nnqz29RGZj8/CyVZNS8P7wYdB0Gs4yOmPz8cyOPxr3czsGdXPrh2bCuvvpIzKG+tPeWuMhghdqD6Meo22P0F/Paquud845u1XuUe9FxFBIXfXwgX3Kpe101ScqsH5Q01XNzTQqXrVd0xsSefbDvBtuN5HMkspldT5tsLIYRoE+TdmhBngIIy9Q1lsJ+RQB8j6YXl/HIgiwsGxHp4ZS0ro0pQnlNiJquogg7B6t5sT2fKtfJ1V7uvg7q39FhWCafyy6p9v2pQblfg+51pzKvSIV0rXfcx6NvdiRd30Rq9OYPy8kLI3A+HVsC+byD3qPr9sC6QcI4jAB8IsQNqbciWa7eQRwiBfr7ueghAlUZvnihfr4vBBEOuhkFXwYHv1ee0JAtKcxwfuVBRiBE7nXXZcPwDePcDdbza1L9DdOs3qtSavBn1Oqx2peFMeQs1easqJsSPhMgADmUUk1FYIUG5EEK0YxKUC3EG0Jq8RQT6MLV/LP9Zd4z/bT/R7oLyqplyULPltQXlnthT3tjydYC4ULURVmodQXmv6CAOZxbzzY7UakG5cxyaZMnrpyhQkg35KWoX9Pxkx+cp3J56hJt8C/H70wR77TUz4n6hMOFRGH6Tune6AYXOEXXu/Zn4erp8vT56PfS7WP04zYe/HeLN77cySH+MOyN/Z0jJRjj6M7wzFvpfCj2nqEF6cAPd15pIy5T37RjC7tSCak3fQP2dfGLZHi4f1pkLB3asHIfWueWCcoCwAPW1lV/WtNGIQggh2gZ5xybEGUALysP8TcweHs9/1h1jzYFMTuSWEh8R4OHVtZz004LyA+mFnN0rinKLjVMFlYGtZzLljZtTDhAXpgblNTPlamn+Xef14t7PdrDzZAFJ2SXOfbdaczLZT14Lux1OboX938H+b9UgvBaRADrA4vgACIyGhLMhcTr0Or/OBmW10brhu/tEiZ/Ri4PyemxIKiKTcFbbh1EYOIXPb4yBVU+qP7c9X6ofOoPaKG7CIxDYso3QtKD8rC5hjqC8eqb8yWV7+PlAJrtO5nNWl3DnibP+LTxqMcxf3Z+eV9rIhoNCCCHaFHnHJsQZIM/ReT00wIee0UGc0yuK9Yez+XBz+5pZrs2B7hTmT2p+mbMDe3JOKYpSeZxHgnLHffo0JlMepu4Tr5opL7fYnAHD6G4RnN0zil8PZfHNjlTumdwbqMyUS+d1B6sZktbBge/gwHIoyax+eXBHCOuqlqKHdYHwrqxJ8+OV3zIZ0y2Mx2cMhPCEJs0J1zi74bv5RIlXlq83wG5X2JyU4/z6VEEZRHSHKz+Ck9vVkvcjq9XGetveg52fqidLYvpBdD+1vD2yl0sVDHXd/4k89XduaJdw3t+UXC0o/+VAJmsOqK+h7GIzz/6wD2i5Jm9VhTsy5QWlkikXQoj2TN6xCXEG0Bq9aVmXuWMSWH84m8+2neDeyb3x92kf+461TPmEPh1YuiXF2ewtKbu42nGeKF+v0Bq9NWZPeXjNTPnxnBLsCoT4GekQ7MslQ+IcQfkp/npeL3Q6XWVW9kzOlCsKnNgCf3ygZlcrqnTj9w2F3lPVsukek9TZ2qcp3pHKXmUHIUSqs7WbqXJufOt1Wq+N1uitwtON3hrhYEYR+aUW9Dq1Z0J6QTk2u4JBr4POw9SPyU+pJ1p+ehzSdsKhH9UPjd6oBuYx/WH0Hep1XJRZVIHZaseg1znL0bVGbxVWGwu+2wtA53B/TuaV8f2uNKDlS9cBwgLU10u+ZMqFEKJdO4PfsQlx5tDe0IU73uBNTIx2vqH8ZkcqV43s4snltZg0R4n6hD7RLN2SwqGMImx2hWNV9pMDVLg5U64oivNEQGO6r3cKq9xTrs1s1vaT94wOQqfTcX7/WPxMu0nKLmHXyQIGx4edOTPKC1OJzd+OfnsGlOVAcToUZUBxhjqarDij8tigGLXsPPEitTFbA1nUYOec8lq6rzeBp/b5V45EazuZ8k1H1Sz52B5RbDqWg9WukFVUQWyoX/UDu50Lt6yFE5shfQ9k7nN87FdPwmTtVz/2fKl2fJ/0OOj9at7habRKlE5h/sSGqMeXWWyUVFj5aHMyx3NK6RDsy2e3jWHyy79S5tga0JJN3jTannIpXxdCiPatnb9jE0JAZZOgUMcbPINex/VjuvL35QdYsvE4V46Ib/MzcEvNVmczrZHdIvA3GSiz2DieU0JSVvWgXNvf7S42u+Isn29M+boWhJRb7OSVWogI9OFwRmVQDmo2fEq/WL7beYpvdpxicHxYZQDYnjPlf3yA8Yf7GWUzQ1Idx5gCoP9lMPQaiB+tNhZzUY3u683kqX3+leXrbSdTvumYGpSP6xlFUnYJqfllnCooqxmUg/oz7TpW/dAoinpSJmOfOnpt9+fqbPT936O74IUG718LyrtEBBDgY8DPpKfcYie7uIIf96QD8NfzetEpzJ/Zwzvz/qZkoGXHoWm0THmBNHoTQoh2zfV3KEKINqtqozfN7OHx+Jn0HEgvYmtSrqeW1mK00vUgXyOh/iZ6x6hB68H0Io7nqEG5lv10957yqnPRG9PozddocHaP10rYj2RVD8oBLhkcB8B3u05hsysUeaipmFtYK+CH++Hbu9DZzBT6dcLe6wIYdiOMfwQuehWu+gRu/hkeOAQz31YDtkYE5ACBvu0jUx7gOLlgtSuUmb0/MLfZFbY4gvIxPSKdfRVOb3ZYL50OQjtD7/Nh1ntw7Vdqz4DCkxg/v4bhSW9CXl1nciqD8viIAHQ6HVFB6u9gWkE5+9LUbRBn91Qby904rht6Hfga9S3e5A0qq5skUy6EEO1bO3zHJoQ4nXNPeUBlUB4W4MOlQzvxydYTfLApmVHdIz21vBahBeUxIeob6MTYEHaeLGDJxuPOku8+McH8npzn9j3lFmtlZr4xI9FA7cCeVVRBan4ZAzqFcjSzZlB+bu8OhAWYyCqqYOPR7Crl6+7dv9zqTu2AZberJcrosI1/hF8K+nDh9IvQm1r2sWoZ7ZbLlHum0VuIn3qSqqDMQnJuCYmxLR84tqT9aYUUllsJ8jUyIC7EMYEgr3FB+el6ngd3bIK1z6NseptO+dtQ3h0DZ10Pk56AgIhqh5+okikHiAzy5WReGRuP5mC22gn1N9E1Ur0sISqQj24ahU6na5Xft1B/x0g0afQmhBDtmmTKhTgDODPlAdXfNF4/JgGAFXvTnfux62Kx2dlxIh+b3b2l367SZpR3dMz2vmZ0FwJ8DGxNynVmmXrHBgPVM9fuUPX+jPrGbRPoXGUsWtX98b2ig53H+Bj1XDiwIwDf7DjV/kaiFaXDikfhvUlqQB4QBVd/hv3s+0HXOv+NaZnyErMNewu85os9lCnX6XTOUXmnb+PwRpsdWfIRCeEYDfoqYwHL67taw3wC4fxnsc5bQ0bwIHR2K/y+CP4zXj3ZU0XKaUF5hyA1MF6zX+1RMKhzaLXtPmN7RjGmR+uc1AwPlEZvQghxJpCgXIgzgHNPuX/15lZ9O4YwslsENrvC0s21z2vWLN6QxMy3N3DXJ3+0SJDS0rRxaNq+00Gdw/jiL2Po6Pg6OtjXWQrqqfJ1H6O+0Xv3nWPR8so4kVuK2WrHz6R3NoHTzBzSCYAVe9LJdnSKbvON3qxmtbv2a4Ng879AsUG/S2D+FrV7eiuqekKjtAX2Yxd5qPs64AzKT2946I20Jm9akBsX2oTy9frEDmRzzwewXvcthHdT59Qvmgrf3Ak/Pwdb36Nn9hpG6A7Q05AO5YXO8vW9p9TS9cGdw1pmLS4I0zLlZRYUxfv+7gohhGgZbfwdmxDCFfkltWfKAW4Ym8DWpFw+2ZrCnZN6Ors1n25/mjpebPnudJ4J2ceTF/XzquZw6c5MeWUzqP5xoSybP47nl+9nbM8o0hzZNndnypsyo1zjzBQWlPFHSh6gZsn1p2Xch3cNd85n/+1wNtDGM+Ul2fDZdZCyUf06fjSMfxB6TnbL3fuZ9M6RXCUV1mY9lxVWm/NEkCd+Js5MuZcH5Vab3dnfYkx3dc921dd/S1K6jIVbf4GvboXDP8GfHzovewHAF/jibwD8JXQk23WXc1jpDKiZcnfR/mZrvSJC2tuWFCGEEIAE5UK0exab3ZmlCw+oOQbq/H4xdAz1I62gnOW707jsLPWNpzaCS5NVVOH8fPGG43QK8+fmc7q38updl+bcU169Q3NMiB+vXTUUgHfWHgXcnylvyjg0TZxzLFo5Kxydnyf26VDjOL1ex4zBcbz761HnSYc2F5QXZ0LKZnW++N6v1Q7aviFw6bvQ50K1gZeb6HQ6gnyNFJZbKa6wEtOM29JK18GzQflxLw/K96UVUlRhJdjPSD9H07QWK1+vjX84zPkM9n8DWQehOJOinFMcPHqUaH0BXXxKwFxMQsFWfvT5nQ9s5/OadRaD48Nafi118DNVdn8vKLVIUC6EEO1UG3vHJoRorMKyyr2IIbWUMxsNeq4d3ZUXVx7k/Y3HueyszqQVlDHnP5sZnhDBS1cMBiCzSH1TfF5iNGsOZPL8jwcYnhDBEDe+Qa1PeqGaSetY29gkBy0odnejN20uemM6r2u0MvXj2SUccHR+njogttZjZw5Vg3KN13dfzzsOSesgZQukbILco9Uvj+gBcz6FDr09sjxnUF7evGZvWpO3AB8Dhkb2FGgJbSVTrpWuj+oW4Xye4hw9InJLzJSZbfj71F7J02R6PfS/1Pnlln0Z3Lz/dwZ0CuH7u86B3CTSvniAjmmrmWdcwUzjRiKOVMCQaxrd0b+pwgN8SCsoJ6/UTLxjn7sQQoj2RfaUC9HO5TuC8mA/I8Y6yqevGhGPj1HPzpMF/JmSx9+XH+B4Tik/7Epz7mPUMuUPXtCH6YM6YrMr3PPpny3Wnbq5tPL1WmcZO/g6gmLPZcqbHpQXlFmosNrpEhFAv461d9BOjA2hT0xlAzivzaopCqx7CV4fAt/eBTs+cgTkOojuD8Nvgsveg9vWeSwghyrN3pr5Gvf03PgER1CeU2J2TmLwRtp88tFVJkGE+BsJdATiDTWjbAmnN3kjohtJk//NteZHOWKPI4JC+PZO+O95cHJ7q68HINRfmr0JIUR7J0G5EO1cXZ3Xq4oM8mXGIHXW9f99vYfvdp4CoMxio7DMitlqd3Ywjw724+8zB9Ix1I/jOaU88/2+Vn4EDTNb7c7mZlr39dpoQbHbR6LZ1BMbTdlTHhZgwr/KPv8LBsTWu5f/kqFxzs+9snzdXAL/uxF+fgZQIH4UnHM/XPM/ePg43LERLnoFBs0G36CGbq1VtdSscuc4NA9VLgT5Gol2zLv31hJ2i83ONm0/eZVO5jqdrnVL2E9TdUa5pkOQL7/ZBzLN/A82dL8HfILh1B/w30nwwwNgad2TBdq2ozwZiyaEEO2WBOVCtHMFjs7rYf4195NXNXdsV0Dd11nVqYIysovVLLnJoCPM30RogIlXZg9Bp4NPt51w7nX2lAxH53Ufo97ZYb02Wvl4hbu7rzejfF0NSiqz/1P71166rrl4cJzjevWfiPGI/BNqp+u9X4PeCBe9Bjf9BOc9Cb2mgH+Yp1dYjXNWubmZQblzHJrnfh7eXsK+J7WAErONUH8TfU+bpR5XZSxgazt9Rjng7L5uwYgy5i64azsMvlq9cNt76qi+4xugLL/hO7CaITcJCk5CWR7YGs5+a7/H3lzlIIQQonm8MI0ihGhJrmTKQR0hNrRLGH+m5BMWYCLEz0RKbinpBeXOIDYqyNfZ9XtMj0huPbc7//71GI9+tYuhXcJqNFlzF+c4tBC/erPInsuUN718HdSg5GhWCTEhvgxtYA9/5/AAXr1yMOUWO2G1NPbzmGO/wv/mQWm2Omf8yg+h61hPr6pegb5qhUJxRfNGojkz5R6sXOjeIZAtSbleOxZNK10f1S2ixmQB7aRUS3dgr02N8nXU8vEOwb6UmW0Mig8FPxNc+g4MnAVf3w6Z+2DJherB/uHqqLWI7hDRTf28LBdO/g7pu9U+CsppryeDrzpH3SfI8W+gWiXi+Hp2gQ8pukTyinu1+uMXQgjhGRKUC9HOaWXn2r7E+jx4fh/u/2In/ze9L8v+TCUlt5S0gnKsjrnkWgms5v4pfdhwJJs9qYU88MVO3r9xZI031O6Q5sJ+cqjMVHtqTnlTuq8DJEQGsv5wNlP7x7r0/F46tHOT7qfFVRRD6u/w22tw7Bf1e7ED4apPICzeo0tzRcvtKVd/Bz25nSAh0rsz5afPJ69Ka/bW2plyRVFqDcr1eh3L5o/DYrVX79PQczLcvhFWPAxJ66EkU81+l+Wp5e11MfqBYgeboxzdVgFlFWrwXouJwERfyN/2BuzxgdJcMPlBUAwERav/BnaAgEgIiFBPDPhHOD53/Gv0rfW2hRBCeAcJyoVo5woc+xBrG4d2urE9o9j06HkAbHZkrtIKylBQg/IOpwXlPkY9r105lIveXM/6w9ks3nicm87u1pLLd0lGLTPKa+MMyt2cKW9O+TrA7RN6EB5g4qazvWcEXTXWCsg+DJn71ayh9m9+cuUxeiMMvQ6mPqdmAtuAoJYKyiu08nXP/ZfrzWPRzFY7vx/PA+oIyt20pzyrqIIKqx29rvI+NZ3C6uhVEdQBLl+kfl5RrGbC85LUEnXtX59A6DQM4oZChz4Q3FHdX2I1g7lY7bPg/Ciu8r1iqCjm6M51dMpYS5g5A7Rt5ZYSKM1Rf89cYQpQA/fxD8PQa5r0/AghhGg9EpQL0c5p3dcbu79Ya5iWVlDuHE/UIbhm0NszOoj/m96PJ5bt4YUVBxjXM5LE2Nq7g7cWZ6a8gfJ5rdGaxaq0+pqqaony9fvO79OSS2q+zAOw/iVI2wU5R2qW5GoCo6HPNDjnPghPcOsSmyuopRq9ebj7Oqjl66BmyhVFqXebh7vtTs2nzGIjPMBE7+jgGpd31MrXWzlTrmXJ48L8m/a76hsEsQPUD1cYfcDoyGTXY7vPpVzy5Wau75LDQxcPVzPhljI1M1+cCcUZ6r9leWoWvSxPzbprnys2sJSqJ8m+vQs6JELnYY1/fEIIIVqNBOVCtHP5jShfr0oLcNMKypyjxE7PlGuuHdWFtQcyWXMgk79+soNv7hyHn6mF5wnXQ5tR7nL5uof2lDel+7pXOvoLfH49VFRpCugbCjH9ILovRDv+7dAXAmtmPtsKZ/f1FppTHuLBTHl8RAB6nbqWrOIKoms5weYpWun66O6RtW7P0LLUpwrKWvWEQm2l694gLMBEMQFsVOKg8/DGXVlR1N/T0lxY/TTsWwZf3gR/+c3j0w2EEEJUkqBciHauMlPeuKZfWnYqraCcAB/1T8Xpe8o1Op2OFy4fxAWvreNgRhH/XHGQJ2f0a8aqGyfNxfJ1Lfvl9j3lzSxf9yp/fAjf3wN2K3QZA+c8oAbgIXFqSW470vLd1z33X66v0UCncH9O5JaRlFXiXUH5sbr3k0PlybZyizqaMSKwdRoYem9Qrj7eJnVf1+nAL1T9mPE6pG5Xy+pXPAyXvN3CKxVCCNFU7eAdohCiPtqe8rBGZsq18vX0gnIyi9SRaHVlykHtzP7i5YMBWLQhiXWHspqy3CbJcDZ6q3tGOVRmqt2+p9wxp7yp5eteQVFgzTPw7Z1qQD7wCrj+G+g1GUI7tbuAHKrOKW9e93XnnnJfz46o6xqhlrCfyGv9LuauqrDaKveTd689KPc1GpxjyVqzhL22GeXeQBvz2Ow55f5hcOm/AR38+RHsXdbcpQkhhGghbfgdohDCFU3dU66Vr5eabRzLKgbqzpRrJiZGc/0Ydd75/V/sJLekmW8iXWCzK2Q4Tho0uKfcqAaObW0kmsfZLPDVLeoecoBzH4TL3mv3HZ2DHCPRWqz7ugcz5VCZcc4obN2GaY2x80QBFVY7UUE+9Iyuu5y602n7yt9bd4yL3lzfon9japtR7g1Cq8wpt9ub2Q8jYZza3wHgu79CQWozVyeEEKIltNF3iEIIV+U53rQ2Nij39zE4MzRFjvLb+jLlmscu7EvP6CCyiip49KtdjVxt42UXV2CzKxj0ugbX52NQgywpX28Euw2+vg12f6F2UL/4LZj0eLvMjJ+uOSPRthzL4e1fjmC3K14xpxwqt3ekuWHet6u0/eSjukfWu1e8sgO7uvYlG4+zJ7WQ9YdbriLHa8vX/dXydUWBwvImlLCfbsKjaif48nz1d9vu3r+HQgghamqD7xCFEK6y2RUKHQF1Y/eUQ81ycFeCcj+TgdevGoLJoGPl3gwOpBc2eJ3m0PaTRwf7OrvE18VTc8orG721sUDWble7Ne/5EvQmuHIpnHWdp1flNoFN7L5+LKuYGxZv48WVB1l7KNO5pzzYw5nyGEclSXpBhUfXUdWq/ekAjK1jP7nGGZQXlFNcYSXVEZwfzy5tkXWUW2xkFKrPi7cF5T5GPYE+6gnFg+lFXPzWb7yx5nDTb9BgglkL1TFpx9fD9sUttFIhhBBNJUG5EO3YvlOVAXFju69D9cZpof4mfI2udVTvHxfKOb06ALBmf2aj77cx0p37yRtuXGVyBMVWu9L8MtBGMLfV8vWfn4EdS0Gnh8sXQp8LPL0itwpuQqbcYrNz72c7KLOo+9B3nSxwBvWeLl/Xfp+1aQWetie1gD2phfgY9Ewb0LHeY7W1p+aXcTijyPn95JyWmbt+Mk8N7oN9jY2uKnIH7aTqsz/sZ9fJAj7ZmtK8G4zsAZOfVj//+Vl1dJoQQgiPaWPvEIUQriq32Ljv8x0ATO0f06SAsGpQ3tB+8tNNTIwG4JcDrR2UO8ahNbCfHKqXj7uz2VubLF/f8yX89or6+SVvQ79LPLseD6gsX3e90dvrqw+z82SB8+s9qQXO7R+enFMO3pcp/3SbGlie3z+mwY7q2li0tPwyDmcUO79/vIWC8qpN3rxphrtGO1GwO1V9baUXllNhbV4DQobfpM4sL8uFX//Z3CUKIYRohjb0DlEI0Rj/+PEAhzOLiQry5blLBzbpNqoF5SGNC8onOYLyP1LynPvaW0NaYWMy5ZV/8tzZ7K3NNXpL3Q7L5qufj/srDLnas+vxEC0oN9vsLgVAW5NyeXvtEQBuHJcAwI4T+VQ4TsoEe7j7uvb7nF1c4fYtHKcrNVv55s9TAMwZ2aXB4yv3lJdzqEqm/HhOy5Svp+R4535yTfhp248UBdLym9mwz2CEC55XP9/6H8g62LzbE0II0WQefYe4bt06ZsyYQVxcHDqdjmXLltV57F/+8hd0Oh2vvfaa29YnRFu1PTmPJRuPA/DiFYOc44Qaq+qe8g6NvI1OYf4kxgZjV+DXVhyPlu7ijHKoHIkG7t1XbrGqpfJenynPTYJld8B/p4C1DHpOhvOe8vSqPEbbxwsNZ8sLyy3c+9kOFAUuH9aZ+8/vg04H2cWVJ6QCfV3b/tFaIgJ9nL8DmUWe7cD+w640iiqsdIkIqHMUWlUdHd3XM4rK2ZdWuS0nt8TctPndp0nJVStuukR6Z1AeWqWkXntdnmyJ0XY9JkGfC9Uxh4svhB0fqxG/EEIIt/LoO8SSkhIGDx7M22+/Xe9xX3/9NZs3byYuLs5NKxOibdualAvA+f1imNgnusm3E1ctU95w0Hs6LVu+phVL2NNdnFEOoNfrMOq1sWju31Pu462Z8vwUtaHbW8PVPeSKTX2jPmsh6D0bSHqS0aDHz6T+zBraV/7ksj2k5pfRJSKApy/uT5CvkW5Rgc7L/U0GjB7++et0OmJC1ZNr2u+Np3y67QQAV46IR99Ag0aAqEBffAx6FAXnXHNNSgtky711RrlGOyl6Tq8oRnaLAOBEXstUCTDtn9ChL5Rmw7LbYcl0yNzfMrcthBDCJR59hzBt2jSeffZZLr300jqPSU1N5a677mLp0qWYTN7XfEUIb6Tts65v7q8rqpaENzZTDpVB+a8HM7G2Url4ula+7uJJA3d2YN95Ip8jmcVVGr152V7VgpPw/b3wxlnwxwdqtqznZLj5Z5jzCfiHeXqFHqftAy8os/D9rlP8mZKH7bQmgd/sSGXZjlMY9DpevXKI8zoDO4U6j/F053WN9nuS7sFZ5RmF5WxPzkOnU6sKXKHX65zZcu33qU9MMNAy+8q9dUa55voxXZkzsgt/v3QgncPVNZ5sqaA8LB7+sh4mL1A7sidvgHfPhlVPgrll9uwLIYSon3e8S6iD3W7nuuuu48EHH6R///4uXaeiooKKisomNoWFapmbxWLBYmmB+Z6iUbTnXJ5799Jm+UYHmZr13EcFVP6JiAgwNvq2BnQMIszfRH6ZheeX72P6wFgGxIXU20ipMa8ZRVGcI9GiAg0uXcfHoKcUGyXlFVgsrXeiL6/UzBX/3oSPQU+fGPXkiF6neM3vgv6PJeh/egydTS2vtncbj/3ch1E6j1QP8JJ11scdf18CHKXCD36xk/3p6l7m8AATlwzuyAPn9ya7uIL/+3oPAPPHd2dQXJBzPX1jg/jGcTtBvq69Plub1rAxNbfEY+tZe0AdgzYwLoQIf9efl9gQX5IdWfGYEF/6xwVzMKOIo5lFdd5GhcWGr6my2qO214yiKKTkqsFnXEjz/ma2lvgwX/42IxGAjqHq/vKUnBb+GY6aD4kXY/jpMfSHfoQNr6Ps/hLb+X9H6XNhy91PGyPvY0RzyOvnzNaYn7tXB+UvvPACRqORu+++2+XrPP/88yxYsKDG93/66ScCArzzDPiZYNWqVZ5ewhnlQIoB0HHy8F6W5+xp1m0FGA2UWnUc27eD5al/Nvr6fYL0bCnTs3BDMgs3JDO+o53LEhrOUrvymimxgNmq/hn7c8NadrtQ+2O3qs/NL7+u41Bgg4c3WUqxujaz1c72lHwADuzdw/Ks3a13py4KLjvJhINPoFNsZAf14UDsLHKCE2FXNuxa7unlNVpr/n2xlauvl/3pRehR8DVAXqmFJZtSWL8nGQUortCREKSQUHaQ5csrm2UVF+gANSC0lpWwfLnnn9vSbD2gZ9POA8QW7PPIGj4/rK4hhvxGPSf2YvV6AOG6MiqyTwAGNu46RLfSA9WOVRT4LkXPL2k6rulhZ3iH6tUNVV8zhWYosxjRobBr06/s89JdJpqsHPV1tftYGsuXn2z5OwicQ0z3RAad/JCAwpMY/3c9e+NmcyTmopa/rzZE3seI5pDXz5mptNT1iiavDcq3b9/O66+/zh9//NGo8SSPPvoo9913n/PrwsJC4uPjOf/88wkJCWmNpYp6WCwWVq1axZQpU2T7gRs9s3stYOaiSePoH9e81/0O3UHWH87m5stGNWmk0wSzlW93prNibwYbjuZwwhLMhReOq/P4xrxm9qcVwe+biAz04eKLzndpPf/cv47C/HJGjhnH4M6hDV+hiX4+mAW7q5/EGDZ0CBcOrn8ec6uz2zB8MB29YsPe83xCZy9llBeOgHKFO/6+fJS2jVTHHuZnZw7g0iEd+flgFg9/tZejReo+80AfAwtvGVOj9LmwzMJb+34BoHNMJBdeOLxV1tgYGRuT+SXtIAGRHbnwwsEuXcdmVzBb7fj7NL+/gN2u8LddvwJm5k4dySjH/mhXHFx9hG2/HgNgbP9uDOocyg8ndmHzj+DCC0dWW+9T3+1jzalUAMxhXbnwwn5A7a+ZP1PyYftWOob6c/FF5zb7Mba2+NQCFh/aQqnOjwsvHN9K93IhWO7F9uvzGLa8Q7+sH+h9xdMQ1PQeJW2VvI8RzSGvnzObVrHtCq8NytevX09mZiZdulSOSrHZbNx///289tprHD9+vNbr+fr64utbc++ryWSSXwYPkufffcxWu7Pjc3xkULOf96cuHtCs64eaTFw3thsTEmM455+/cCKvFL3BiKGB5k6uvGayS9WyoNhQP5cfp49RDSwUnb5VX5P5ZTUbg/n7esHvweaFkPo7+ASjn/Eaep/650O3Ba3596VndDDbjudxz+ReXD06AYDpgzvTKzaUeUu2cTKvjAWXDKBHTM0TPJEmE10jA0jOKSXY3wt+9kCncLU8JLPI7NJ6LDY7cxdtYU9qAWsfnEiH4KZNctDsPVVATomZAB8DI7t3wNSIiQTxkZWlLYkdQ+kRrZ5wTM4tdT4Wi83Og//byXc7TzmPLSy31nisVV8zpwrVv5ddIgO84mfUkG4d1MedWVSBDT1+plZqxmgKVUemndiC7tQfmDa/CdP+0Tr31QbI+xjRHPL6OTM15mfutUVa1113Hbt27WLHjh3Oj7i4OB588EFWrlzp6eUJ4bUyHA2cfIx6IgK9J+CKC/PHx6jHYlNIbYlRPuDcT+7KODSN1gG9tRu9ZRWpvS2qZk893n09LxnWOLb3TFkAoZ08u5424KkZ/fjp3nO5Z3Lvat/vHRPMqnvHs+b+8fU2KxvgaPYW7Ocdb8a05o1pLnZff231IbYk5VJitnEg3fUz/nVZfzgbgNHdIxs9IrDq73mvmCC6Rqm/W9nFZorKLZRbbNz24Xa+23kKk0HHjMHqxJbcEnOtt6dJ8fImb6cLCzA5x6Kl5rfM39I66XRw3pPq578vVCc1CCGEaHEefYdYXFzsDLgBkpKS2LFjBykpKURGRjJgwIBqHyaTidjYWPr06ePJZQvhEesPZ7FiT1qDx1UNVBuz9aO1GfQ6EhwzgI9lF7fIbWY4x6G5HpSbjOpzYm6lbvAaLSifMbgjY3tEotNB9w6tuIm9IYoC398DllLoOg6G3ei5tbQhfiYDvR1dvk/n72OgR4f6Jxyc0zMKoMHj3EX7XcksKsdur38s4NakXP619qjz65aYB77+cBagjvZqrE5hlWMPe8UEE+JnItJx4nHfqUJuWLyVnw9k4mvU85/rhzN7uHqyJL+0/nW3taBcp9NV6cDeykE5QPcJkHAO2Mzw6wutf39CCHEG8mj5+u+//87EiROdX2t7wefOncuSJUs8tCohvI/drvCXD7dTarGx/qGJzjdktUlzjENzdUSYO3WLCuRQRjFJ2SVMaIFza5UnIBqeUa5xW6a8WA3KOwT5suiGEeSUmKsFFW6381M4+jMYfGHGG6D32kKpdmX28HiGJ0TQPcqDJ2SqiA72RacDi00hp8RcZzl6QamFez/bgVIlbm9uUF5mtrHNsT//nF4dGn397h2COC8xmo5hfs7+Fl0jA8gpMXPrh9spKLMQ5Gtk4dzhjOoeyZ7UAkCdhFAfb59RXpv4CH8OZhQ5R7m1Kp0OznsKFk6GHR/D2L9Ch94NX08IIYTLPBqUT5gwAUWp/0x9VXXtIxeivSsos1BitgFq9qr+oLzxJd3u0i0qCMjgeHbLzL7VZi3HNOIEhLvmlGuZ8g7BfviZDJ4NyIszYcUj6ucTHoGonp5byxlGr9fRM9o7suQAJoOeqCBfsooqyCgsrzUoVxSFB/63k9T8MrpEBNCvYwgr9qZTWEufhMb4IyUPs9VOx1A/ejShasSg17HwhhHVvpcQFcgfKfkUlFkIDzDx/ryRDOocBkC4I4ueV2pGUZQ6K4e8fUZ5bdyaKQeIHwF9psPBH+CX52D2++65XyGEOENIqkSINiCnyp5ILdNUl3QtKPdkEFiHblFa+XrLBOVNOQFhcmTKLW4qX29uY6xmUxT47h4oz4fYgTD2Ls+uR3hcxwb2lf93fRKr9mXgY9Dz9tVnEef4W9LcTLn2O9GjQ1CLba3RthZEB/vy+W1jnAE5qPPkQa0K0E5qaqw2O4cziigotThP7rWtoFz9mZzMc0OmXDPpcUAH+5bBqR3uu18hhDgDeG33dSFEpaqNin4/nlvvsVr5uvdmyiGphYLypuwp93V7ptzDQfnOT9Xslt4EM98Fg3c0HBOeo1aWFDiD0ap+P57LP1aoM7+fmNGPgZ1D+flAJtD8oLzUERi3ZLfwa0Z1wc+o5/z+sc6TBxp/kwFfo54Kq528EnO1kY7vb07hHysOYTLoUBR1rJ03NcZsiJYpP+GuTDlATD8YNBt2fQY/PwPXfum++xZCiHZOMuVCtAG5JRXOzw9nFpNXTzfh9Cbss3aXbo59tan5ZZRbbA0cXb+icgtFFWo5bWP2z7sjU15qtjozcx4NygtOwo8PqZ9PfBRimzfeTrQP2gm79ILqAV1OcQV3fvwnNrvCxYPjuHaUOpI0xF8NZgubu6fc8Tsf0ALzzjXBfiZuGNetRkAOakO08AA10D692dv25HxAzaID9I4N9qrGmA3RMuXJOSU8sWwPk1/51bmHvlVNeAT0RjiyGo5vaP37E0KIM4QE5UK0ATmnBeG/J9ddwn7Ki/eURwX5EOxrRFFodoMibfRbsJ+RQF/Xi360PeUVTciU70kt4O/L91NSUf/e2uwi9eflbzI4Rxe5nd0O38yHikLoPEJtziQElZUlhzIqpyDY7Qr3fr6T9MJyuncI5O+XDXQGqaH+anVFYXlzG72pvzf+rTVXuxZhjhL23NOavWl7sZ+8qB//uGwgL10x2G1raglaU7r8Ugsfbk7mSGYx/1p7pPXvOKI7nDVX/XzN36ARfYGEEELUTYJyIdqA3OLTgvI6StjNVjvZjq7fjSnpdhedTkc3R4On5u4rb2pDu8pMeePfTL7582H+s+4Yn/9+ot7jsorVtUUF+3gu+/b7Qji2Foz+jrJ12a0kVBN6RwOwen8GhzOKAHj7lyOsO5SFn0nPv645q1qptxaUN7v7uiNT7u/GE1WVmfLTgnLHfO+ze0Vx1cguXjOyzlWh/iaiHVU42qjF1fszKWrmiROXnPug+nflxGY4/FPr358QQpwBJCgXog3QMuVamfbWOoLyzKJyFEUd+xXppfsjEyLVN5DN3Vee7txP3rgy/eZ0X9e6T/+Rkl/vcc795EEeKl3POQqrnlQ/n7JAuq2LavrFhTC1fwyKAq+vOczGI9m8uvoQAM9cMoDE2JBqx7dUUK7tKXdnUK7tE6+65afUCkXl6u+yR6ciNNOiG0bwn+uGsere8fSMDsJstbNiT3rr33FIRxh1q/r5mmfUqhwhhBDNIkG5EG2A1uhtav8YQC2jLjPX3JOdVqXxmbfuj9T2lSdltUxQ3rGR89h9mrGnXMv0/VHP9gHwYJM3Szns+hw+mQOWUuh2Loy4xb1rEG3CPZPVOdM/7E5j/sd/YFfgimGduWJ4fI1jWyxT7vibFeCR8vXKtec6WnREBvo0auuLtxnQKZTz+8di0Ou4ZHAcAN/sOOWeOx93D/iGQMZu2PuVe+5TCCHaMQnKhWgDtKB8UOcwYkP8sNgUPth0vEZgntaEbuTuppVaJuU0s3xdm1HeyMfqzJQ3JSh3PN+p+WVk1tK5WuP2oDzrEKx4DF5JhK9ugeyD4B8Ol/wL9PJnXtTUt2MIFw6MRVEgr9RCYmwwf7uk9kaAIdqe8jILSjP2EHtL+XpOuXrCUmuW1h5cMqQTABuPZtf7t6nFBERUjldc/4rsLRdCiGaSd2tCtAFa+XpEkA9je0YC8PyPBxjx3Goe/t8ufj+ei6Iozm7KcV4clDsz5S1Uvt7YPeVaprwp5etlVTrG11fCnlWsla+34s/BWgG7voDFF8LbI2Dz21CWB6HxMPH/4I7NEFYz6ymE5q/n9cag1xHoY+Dta86qM1jWMuV2BYobaHJYH0+Ur4dr5eu1ZMo7t6G55A3pEhnAWV3CsCvw3a4099zpyFvAJwgy98LhVe65TyGEaKfabt2WEGcQbSRaVKAvT13Un64RgXz5x0lSckv57PcTfPb7CRIiA5ylmI3dZ+1OCY6gPKuoghO5pc4uwo2V3sSqAK3RW5My5VWC8j9T8rhgQGytx7V6prwsTw3GM/epX+v00PsCGHYj9DwP9B7q+C7alD6xwXwzfxwBPga619PozM9kwMeox2y1U1BmIdivabPuy1thJFpDwh3l61Uz5bkV7S9TDmq2/I+UfL7ZkcpNZ3dr/Tv0D4dhN8Cmt2DDa9D7/Na/TyGEaKckUy6El1MUxVm+HhHkQ2iAib9O7sXaBybw2a2juXxYZwJ8DBzPKWXvqULAO8ehaUL8TJzTKwqA99Yfa/LtpBc2MVPejEZv5eaqQXl+ncdpQXlUUCs027NZ4PPr1YA8IBImPAr37IE5n6hviiUgF40woFNovQG5piX2lTsz5W7cU66Vr+eWVA3K1X87h7efTDnA9EEdMeh17DpZwLGs4oav0BLGzAe9CZI3wImt7rlPIYRohyQoF8LLFVVYneO7qnZU1+t1jOoeyUtXDGbb/03mpSsGM6pbBF0jA5iUGO2p5brk9vE9APhs2wlnANsY5Rab8012bCMbvZkMapasOY3eAHal5td5G9mOEXYtnilXFPjhPkhap5aNXv8NTHgEQju17P0IcZoQP7UKpzlBeZmzfN19RXphzkx55bpzHJny+HaWKY8K8nWe8HRbw7eQOBh8pfr5qqegvMA99yuEEO2MBOVCeDltRnmAjwG/OjJMgb5GLh/Wmc9uG8OvD05sckm4u4zpEcng+DAqrHYWb0hq9PUzHFlyP5PemcFzlW8TM+UWmx2rXXHeRrnFzoG0ohrHKYrSeuXrm96CPz5Qy9VnLYTYgS17+0LUIdTZ7K3pe8qdjd7cmCl3jkRzlK8rikKuow9ae8uUA1wyROvCntqspnyNMu4eMPhCykZ492w4sc099yuEEO2IBOVCeDlnkzcvnTveFDqdjjsmqNnyDzclU1jeuOxbZZM3/0aPftPK1xubKS+tUro+PCEcgD9Sao5GKyyzOverR7XknPIDP8BPT6ifn/8c9Lmg5W5biAaEVunA3lTOkWhu3FMe5ihfLzXbKLfYyC+zUGFvn3vKAc7vF4u/Sd3OtOukm7LWUb3ghh8grAvkp8CiqbD+ZZlfLoQQjSBBuRBeTivTjmxHQTnAlL4x9IoOoqjCytLNKY26rrafvLGl61DZ6K2ikZlyrUmVQa9jZILaAf+11Yd49vt9HEyvzJhnFatrC/Ez1lnZ0GhpO+HLmwEFhs+D0be3zO0K4aKW2VOuZtlb7PfCBSF+Rgx6NQjPL7WQmqf+fnYI8nHrOtwl0NfIlH4xACzbkeq+O44fAX/5DfpfBooN1vwNPpwJhW7qBC+EEG2cBOVCeDmt83p7ypSDuif+L4695Qt/S3IGva5ozjz2pjZ6K6vSpOqiwR2JCfElr9TCf39LYupr65jx5m+8v/E4hzLUBkstVrpeeAo+vgospdB9Ikz7JzSyOkCI5gppgaC83KL+zrkzU67T6Zwd2PNKzZzIKwWgUzvMkmu0EvbvdqZhbULvjCbzC4XLF8Elb4MpAJJ+hXfHwcEV7luDEEK0URKUC+HlKsvXW2m8lgddPCSOTmH+ZBdX8MX2ky5fr6nj0KAyU97Y8nVtP6yfyUCPDkH89vAk/nv9cKb2j8Go17E7tYCnvt3LHUv/AFooKDeXwCdXQdEpiOoDVywBQ9PGUQnRHM7y9XILiqLw0eZktifX3L5RF6vN7tzW4c495VBZwp5XaiY1X/3b0Sms/Qbl5/buQHiAieziCjYdy3Hvnet0MPRauG2d2vOiNAc+uVItZxdCCFEnCcqF8HJao7fI1hiv5WEmg55bzlHn6f5n3VGXszqVe8qbkSlvYlDu76Ne32TQM7lfDP++bjhbHjuPp2b0o1/HEOfxXZrbbM9uh69uVUvXAyLh6s/AP6x5tylEE1UtX//zRD6PL9vDo1/tcvn6VScX+LsxUw6Vs8rzSiyk5pcB7a/zelUmg54LB3YEYNmfburCfrqoXnDzGhjl2Gqz9h9QnOWZtQghRBsgQbkQXi63HTZ6q+rKEV2IDPThRG4Z3+9ybf9hmmNPeUwT9pT7aJlya+M6E2szygNMNcc5RQb5cuO4biz/6zn8cPfZPD2jH/dN6dPotVWzZgEc+B4MPnDVxxDRrXm3J0QzVC1f33uqEKjcRuIKbfuHTlc5AcFdqmbKT+SpQXl7zpQDzByqjklcuTe9UVuDWpTRF6b9AzoNA5sZ/ljimXUIIUQbIEG5EF4uu50H5f4+Bm4clwDAO2uPYrc3HCznOZ6TqCZUDzQ1U651X/drIMvXPy6UG8Z1a1JpvdOfH8GG19TPL3kbuoxu+m0J0QJC/CqD8sMZamPDonKry9Utpc6TWoZGT0xorghHUJ5fauZErhqUt8fO61UN6xJOpzB/iiusrNmf6dnFjLxN/XfbQrA1vSeBEEK0ZxKUC+HltEZv7a37elXXjUkgyNfIwYwi1h7ObvD4fMfM4VD/JgTlhiY2enPOWG7FP5u5x+Dr2+Hbu9Svz30IBs1uvfsTwkVVR6JVnTaQ72Ljt8rtHzUrTVpbWKC69tX7MzmWXYJep9AnJsjt63AnvV7HxY6Gb27twl6b/jMhMBqK0mD/t55dixBCeCkJyoXwcpV7yttfozdNqL+Ja0Z1AeDf65JQ6kmW2+wKRRXqaKWwgMY3PdMavTV5T3lrNKnKS4Zv7oQ3h8POj0Gxw1nXw8THWv6+hGiCyj3lVg5nFju/r1WtNETLlGs9Gdwp3JEp33EiH4BRHZSWm47gxWYOUUvY1x7MpKDUgxlqo686yhFgy789tw4hhPBiEpQL4cUURXF2X2/PmXKAm87uho9Rzx8p+Rwtqvu4onKLM2jXAoXGaOpItHJnpq8Fg/L8E/DdX+HNs+DPD9X5vj0nw80/w8Vvyugz4TVCHSfAckoqnH0uAPJcDPa035/aejK0Nq18HcCo1zGlkxvHhHlQn9hgEmODsdgUlu/x8Lzw4TeC3gQntsDRXzy7FiGE8EISlAvhxUrNNiocwWN73VOuiQ7x4/JhnQFYnVr3n6Z8RxAQ6GNwZr0bw6epI9Gcc8pbIKiw2+Gnx+GNobB9Cdit6gzym1bBtV9C52HNvw8hWlCIn/q6P72KJa+0cZnyhnoytIaqFTWXD+tEZDPaPbQ1lziy5cv+9HAJe3BsZbZ8+QNgrfDseoQQwstIUC6EF9MyUr5GPQEeeDPrbred2x29Dvbn650dnk+n7WENC2jaSYqmZspbtPx29+ew8U2wW6DbuXDjCrh+GcSPbP5tC9EKgnyNGPQ1KzdcLV8vs1Q2enO3KEepusmg4/Zzz6wpBtq+8q3HcznlGAfnMZP+T91bnnMENrzh2bUIIYSXkaBcCC9WtXTd3R2LPaFrZCAXDogF4D/rk2o9psARlDeldB3UN+YAVrviUqd3TXlL7Sm3muGXv6ufT3gM5n4HXcc07zaFaGU6nc6ZLa/K1fL1MrPaB8LdM8oBhnQO4+azu/HSFYOJa+ej0E7XKcyfkd0iUBT4bqeHZpZr/EJhquNv3/qXILf2v/FCCHEmkqBcCC+mdV6PaMLor7bq1nPUTNaKvRkkZZfUuLyy83rTgnKfKjOSG9PsrcUavf3xPuQnQ1AMjL2rebclhBtV/Z3TGqXlu1i+7tz+4YGgXK/X8fhF/Zyl3GeaS5xd2D0clAMMvFytDrKWw/f31NwPIYQQZygJyoXwYpmFalAeHXzmbILs2zGYfmF27Ar8Z93RGpcXOMvXmx+UN2ZfeVlL7Ik1l8Cv/1Q/H/8Q+AQ0/baEcLOQKkH5yG4RQCP2lLfm9AJRr+kDO2Iy6NifVsihjHq6aLqDTgcXvQZGPzi2FnYs9ex6hBDCS0hQLoQXy3AE5TEh7X98T1WTHd2Rv9yeSkZhebXLtEZvTQ3KTfoqmfJG7CtvkUz51v9ASSaEJ8DQ65t+O0J4gJYp1+tgWJdwAHJLXOy+7jipdSb0xvA2YQE+jO8dDcA3np5ZDhDZo3Lc48rHoCjds+sRQggvIEG5EF4so0gNSM+kTDlAjxDo1zEYs83OlqTcapdpQXmof9NK+vV6nXNfucXW+D3lTQ4qKoormxtNeBSMZ86WBNE+aJnyrpGBdAxV/ya5Wr7ubJQomXKPmNJPDcp3nSzw8EocRs+HuKFQXgDf3ytl7EKIM54E5UJ4sUxHljgm5MwKygE6h6sNmU5/09/cRm+Ac5RaYzLlzpFOTQ0qtr0HZbkQ0R0GXN602xDCg7TfuV7RQc7pB66WrzsrTSRT7hGBvmqTvsZOnWg1BiNc8rY6u/zgciljF0Kc8SQoF8KLnanl6wDhjvL0/NO6OxeUqUFAU8vXocpYNJvN5es0q3y9olgdgQZw7kPqG1Ih2phe0UGAup88PFD9/XO9+7qUr3uS0bFtx9qIiROtLqY/THpc/fzHhyHvuEeXI4QQniRBuRBeLOMMzpRrWbnTM3HOPeUtkil3/Q1qs7pHb/svlOaoWfKBVzT++kJ4gbljElh5z7ncOK4bEY5MeX6p2aXRglK+7lnOUZCNaG7pFmPvgi5jwVwMX9wAqX94ekVCCOEREpQL4aWsNjvZxVr39TMvU65lwgtOy8Tla+XrzcmUa0F5I96gNnlO+akdsPYf6ufnPCBZctFm6fU6+sQGY9DrnOXrdgWKyq0NXreyfF1e/55gdPzNa0wfDbfQG+DSd8AnGE79Ce9NhIVTYe8ysDX8uhJCiPZCgnIhvMgLKw5wx9LtWGx2ckrM2BW103Fk0BkYlDeYKW96ozRfo/YGtQnd1xuTKS/Ogk+vAWsZ9JwCg69q1DqF8FY+Rj2Bjt8FV/aVl0mm3KNMekem3O5lmXJQp1HcvBoGXaXuMT+xGb6YC28MgQ2vQ1mep1cohBCtToJyIbxEWkEZ76w9yvLd6ew6WeCcUd4h2BeD4w3VmUQLurXMOICiKBS2QKa8OY3eXA4qrBXw+fVQeBIie8Ks/6pZISHaifBA9Xc015WgvLnTC0SzaJlyq7dlyjXRiXDZv+HePWrfjYBIKDgBq56EV/rBmmekQ7sQol2ToFwIL/HT3gzn50czi8/o/eQAoQFqmWvVRm9lFpuz5Lw5e8orG701oXzdlaDCbodlt0PKRvANgas+Af+wpixVCK8VXmVfeUNKzWopcpOnF4hmMWpjIL0xU15VcCxM+j+4dx9c/BZE9wdLKax/CXZ87OnVCSFEq5GgXAgvsWJPuvPzI1nFZ+yMck24f803/FqAbjLompVx05oeuZopt9jszr2YLmXKVz0Be74EvRFmfwAdejd5rUJ4K63vQ15Jwx3Yyy3q75pkyj3DpPfyTPnpTH5w1nVw+waY8Jj6vRWPQMFJz65LCCFaiQTlQniBnOIKtiTlOL8+kll8Ro9Dg8ry9IIyi7O7sxaUh/r7oNM1vaTfmSl3MSjXsuRQT6YvLxnWvwL/Ggub3lK/d8m/oMfEJq9TCG8WEej6rHItUy5zyj3DmSlvK0G5RqeDc+6HTsOhohC+vUvK2IUQ7ZK0QRXCC6zen4FdURuQVVjtHM4scnZcP2PL1x3l6Vp359AAEwXafnL/5v3pMhka1+hN2w+r01U2iQOgJAf2fgW7/6c2J9IYfGDKMzD4ymatUwhvppWvu9ToranTC0SLcI5E8/by9doYjHDpu/Du2XD0Z/jyZpjxOvgGeXplQgjRYiQoF8ILaKXrc0Z2YcnG45zMKyM5pxQ4czPlvkY9AT4GSs028krNjqBcffOvjWNqzm1DIzLlZkfprcmgZugL0+CH++DwT2DXxvbooNs56hzyvjPAP7xZaxTC2znL10vrL1+32xVn+bpkyj3D2NbK108X1QsuehW+uRP2/A/Sd6tbg6ITPb0yIYRoERKUC+FhheUWfjuSDcA1o7qwbEcq+aUW/khRx8CcqXvKQc3ElZrLnB3YK8ehNb3JG1SWr7uaKS+1VCm9LS+ApZdDxh71wo5D1EB8wCwI6disdQnRlmjl6w01eiu3Vm7/kD3lnlFZvt4GM+WaIVdDeDf44gbIPqjONJ/xBgy6wtMrE0KIZpM95UJ42C8HMrHYFHp0CKRXTDA9O6gleRWOLG70GZoph8oSdq08Nr8FxqFBZfl6hYuZcm3GcpDRDp9dqwbkQTHwlw1w268w9k4JyMUZR6tYyS2pPyjXxgkC+BklKPcE7W+e1d5GM+WarmPgL+uh27lqV/avboYVj8o+cyFEmydBuRAe9uNutXT9ggGxAPSMrr5P7kzdUw4QHuho9lZaPVMe2txMueMN6oo96fxn3VFW7k3nYHpRtYZuVZVZbOix85T1TUhaBz5BcM0XEDugWesQoi0Ld5wcy2+gfF07qeVn0qPXN71Bo2g6o+N5t9kVlLYewAZFw3XL4NwH1a83/0v9EEKINkzK14XwoDKzjbWHMgGYNkDNtFYNyo16HRHN3D/dloWdNhZNa/Smfb+pYkPVEx2/J+fxe3Jetcs6hvrRNTKAhMhAEqICuWRIHOVmCy8Y/8NE63rQm9S9jB0HN2sNQrR1rjZ605q8BfjIWw5PMRoqczAWm4KPsY2fHNEbYNLjEBAFKx6Gn56AmAHQfbynVyaEEE0i/0MK4UG/Hsqi3GKnU5g//eNCgOpBeXSw7xmdWTq9kVRlo7fmZcrnT+xJjw5BHM0q5nhOKck5JSRll1BUbiWtoJy0gnI2H8sFYOvRTJ4yLGaScR029BguXwQ9z2vW/QvRHoRrI9FKLCiKUueYQq18XTqve47WfR3UDuw+7aVQctRtkLYDdn6i7jU//1m1x4fxzD2Z3Vhv/PEG0QHRDIkeQq+wXhj08nsqhCdIUC6EB63YkwaopevaG9pqQfkZXLoOlcF3wemN3poZlPuZDMwc2qna9xRFIb/UwvGcEpJzStmenMf/Nh9k7skn6Wrfil3R8V6Hh/lLv4ubdd9CtBda+brZZqfUbCPQt/a3FFr5unRe9xyt+zq0wVnl9dHp1K7sWQfh1B/wzR3w87Mw5g44ay74hXh6hf/P3n2Ht1WefRz/anmveMZOnL13ApkESCAJhL1ayqalpYzSQQeFUlZLy2jLKOtlFMqmrLACJMyEhOy94+x4JN57SNZ5/ziSbMd2Io9Ysv37cPmyLB0dPZYfHN/nvp/7CWoVzgr+s+k/1Brm/6OhtlASwxOJD4snISyB+PB44sPMj+SIZEYljiItMq3ZC3Ai0noKykUCpMbl5sut3tL1nr7702LDCXfYqHTWdtvt0LyOLI/1BuUxbVxT3hSLxUKPyBB6RIYwPt7F6TVr+GHIPxnt3ovLGsKvq35Obdzsdn9dkc4q3GEjxG6lxuWmoLym+aDcu3uBMuUB0yBT3pk7sDfFEQ7XfAQrn4dlT0NpFiy4E759GCZeB5NvgOiUQI8yKDndTn4+5uesy13H+tz1lDvLySzLJLMss9nnJIYnMiZxDGOTxzI2aSwjE0YSZu/eCQSR9qCgXCRAlu7Ko7TaRVJ0KBP61O1pbbVaGJgcyabMkm7d5A3qGroV+crX22dLtEbcbshaa+47vnMBZK0hGhhthQIjijf7PsjHW3twkYIKER+LxUKf+AgyDpexcMshfjK9f5PHVdZoj/JAs1gs2KwWat1G5+/A3pTQKJj+a5hyI2x4C5Y8Dvk74bt/wfdPwoSrYM794Oje/6YeKTY0lhvH3QhArbuWrLIs8qvyKagqaPhRWcCB0gNsK9hGXmUeXx34iq8OfAWA3WJnRMIIZvaZyey+s+kT3UeZdJFWUFAuEiCfbTK7rp8xMqXRuvFhPWPYlFlCr7jwQAwtaHgz5Y0avbVH87uKAtj1FexcCBlfQEVew8dTx/J6wTCeLJ5GYll/oIgwBRUiDVw3vT+3v7eRp7/dxWWT+jQZeFfUKFMeDOyeoLxT71V+LPZQmHA1jLsStn8C3z0KmavMLHp5HlzyIli7yHr6dmaz2kiPSSc9Jr3ZY6pcVWzJ38KG3A2sz13P+tz15FbmsiFvAxvyNvDYmscIt4eTFplGWpT50SuqF2lRafSO7s3QHkOxWxV6iDRF/2eIBECt22DBlkMAnDmy8f7Wvzp9MOk9Irh0YvP/OHYH3rXjRZVOnLVuyqrNP+7bnClf9gx8fgcY9bZAC42BgTNh8BwYNAuie/LtK6vILD5E4aFSQEGFyJEuOaE3T36dwcHCSl5bvo+fnjyg0TFVvu7r+v8nkBw2K9UuN66utKa8OVYrDD8Xhp0DOz6Dt66CLfNgYW844/5Aj67TCrOHMSFlAhNSJgBmL5bs8myWZC3hi31fsCJ7BZWuSnYV72JX8a5Gz48OiWZ6r+nM6D2Dk3qdRGxobEd/CyJBS0G5SACs3FtAQXkNcREOJg+Ib/R4enwEv5o1OAAjCy7ejHhheY0vSw5tXFO+eZ65hQ5A0jAzCB88B/pMAVvD8/ZNiATqukcrqBBpyGGz8svTBvOHdzfwzLe7uHxyn0Zbn6n7enCwe9aVu9xdOFN+JIsFhs6FC56G934K3z8BUSlw0i8DPbIuwWKxkBaVxg+G/IAfDPkBNbU1ZJdnk1mWSVZZFlllWb7bu4t3U1JTwqd7PuXTPZ9is9gYnTiaYfHDGBo/lKE9hjKoxyDC7d27QlC6LwXlIgHgLV2fNTwFh02ldM3xdncuqXKxPcfMVidHh2Jr7TZxB1fB+z83b0+6HuY+ZP7R1ow+8RENvg5TUCHSyIUTevHE1xnsL6jgle/38fNTBzZ43LtPudaUB5a3A3uX6r7urzE/gJKD8MU9sPDPYLWbHdqlXYXYQugb05e+MX0bPVbrrmVD3ga+OfANiw4uIqMog3W561iXu853jNViJT4snrjQOGJDY4kLjfN99AjrQe+o3vSP6096dDoOa/s3fBUJJAXlIh3M7TZ8QfmZI3se4+juLbZeRvzzzeZ7NnlAQutO5qyC/10DrioYfAac8fejBuQAfRMaBuXK9Ik05rBZ+eXpg/nd2+v5v0W7uXJK3wad2CtVaRIUvB3Yu0X5elNO+jXUVMCih+Dz28FihSk3BHpU3YbNamN88njGJ4/nNyf8hoOlB1l7eC07CnewvWA72wu3U1BVQF5lHnmVeUc9l91qp19MP/rH9mdg3EAGxA5gQOwA+sX2I9TWvXetkc4roEH5okWLePjhh1m9ejXZ2dm8//77XHDBBb7H77nnHt58800OHDhASEgIJ5xwAvfffz+TJ08O3KBF2mhDZjE5JVVEhtiYPjgx0MMJanablegwO6VVLt+FjKmtDcpXv2hmSmJ6wSUvgO3Yv/76xkc2+FqZPpGmXTAujSe/zmBPXjn//X4vN80Y5HvMlynXRa2A8pavO7tT+Xp9FgvMvAMMNyz+h7mMyWKFydcHemTdUu/o3vSO7u372jAM8qvyyavMo7CqkOLqYoqqi3wfBVUF7C/Zz+7i3VS6KskoyiCjKIOF+xb6zmG1WOkR2oNwezhh9jDf5zBbGGH2MHqE9qB3dG96RfXyvX5MiPayl+AQ0KC8vLycsWPH8pOf/ISLLrqo0eNDhgzhiSeeYMCAAVRWVvLII48wZ84cMjIySEpKCsCIRdru003ZAMwclqxyaD/ERTgorXJxuLQagKkDWxGU15TD4n+at0/9A4RG+/W0tLgw3zZCoKBCpDl2m5VbThvErf9bz7OLdnPVlL5Eh5mVLuXV3vJ1FecFksNTvt5tM+VgBuan3WkG5t/9Cz79vXnfpJ8FemTdnsViITE8kcTwoycr3IabQ+WHzGZyRbvYU7yHXUVmY7nSmlLyq/Jb9Lr9Y/szqeckxiePZ0iPIfSL6YfDptJ46XgB/Rdy7ty5zJ07t9nHL7/88gZf/+tf/+KFF15gw4YNnH766cd7eCLHxdfbDgNw5iiVrvujR0QIBwoqAUiJCaXfESXlfln+DJTnQo/+MO4Kv59mt1npFRfO/oIKQJlykaM5b2waT3ydwe7ccv67dC+/OG0wtW6DZbvNP5J791ADp0DyNXrrylui+cNigdPvMnffWPIYzP8dxPc3d92QoGe1WEmNSiU1KpXpvab77vdm2vMr86mqraLKVUWlq9L3udJVSV5lHgfLDpJZlsnB0oMUVBWwp3gPe4r38Nb2t3zninZEExMa41vP7r3dM7InE5InMDJhpAJ3aXed5rJ1TU0Nzz77LLGxsYwdO7bZ46qrq6murvZ9XVJSAoDT6cTpdDb3NDlOvO+53vs6h0qqABiYEK73pQlHzpmYsLpfU5P7xeNyuVp2wqpi7EsewwK4Tv49hhtw+/++p/eoC8odVkM/syCi3y/B5+ZTB/Dbdzby7KLdXD6xF+sOFJNZVElMmJ0Zg+MD/rPqznPG5umhUVWjv4cAOPVOrBVF2Nb+F2PeTbiu/w7CezQ6rDvPmc4m1h5LbLT/26wVVxez5vAaVh5aybbCbews3Em5q5xSZymlzlIyyzKbfF6YLYxRiaOYkDSBUQmjGNJjCEnhSVia6FOj+dO9teTnHvRB+ccff8yPfvQjKioqSE1NZeHChSQmNl/a8ve//51777230f0LFiwgIqIVGTZpFwsXLjz2Qd2AYUBJpQ2wsHLpInaEBHpEwcs7ZyoKrYBZdhlRdpD58w+06DzDst9laFUxJWG9+Hp/BByY37KBlNW9/rpVKyjd0bKny/Gn3y/Bw2pASriNQ5Uu/vTfL8iqsABWxsXV8NXCzwM9PJ/uOGfKS81/e75fsZLSnd24hL0em3s6p4YuILosm0P/uZLV/W9u9tjuOGe6i1Ge/4xIgwqjggqjgkqjsu6227yd585jr2sv5bXlrDq0ilWHVvnOEWGJIMWWQoo1hXBLOGGWMN9HqCWUg58ebPC1w6JMe3dQUVHh97EWwzCC4jezxWJp1OgNzHXn2dnZ5OXl8dxzz/HVV1+xfPlykpOTmzxPU5ny9PR08vLyiIlRM4eO5nQ6WbhwIbNnz8bh0C+gsmoX4//6FQAb/ny6yqGbcOScuffjrby63AzEv/zN9EbblB1VRT72JydgqSnHdfGLGMPObfF4nv9uLw9+bkbi826cwsg0/R4JFvr9Epw+2ZjDr/+3gahQO1XOWlxug49unsqwnv71cjieuvOcufS5FazZX8STl41lzoiUQA8naFiy1mB7aS4Wo5baST/HffJtEFb3e747zxlpzDAM9pTsYc3hNaw9vJbtRdvZV7KPWqO2RefpF9OPE5JPYHzSeEYmjCQ9Oh2rRVvkdjUlJSUkJiZSXFx8zDg06DPlkZGRDBo0iEGDBjFlyhQGDx7MCy+8wO23397k8aGhoYSGNt4OweFw6JdpAOn9N1WWm6XXdquF6IjQJkudxOSdM/FRYQD0igtnQHJMy96z5U+YTd56jsE+6sJjboHWlP5JdYFEVHio5nEQ0u+X4HLeuN489e1udhwqA2Bsehyj0+MDPKqGuuOccdjMP/gNi7Xbfe9H1Xey2ZX9q79gW/F/2Da9AzNuhxOuhXrrhrvjnJGmDU0cytDEoVw24jIAqmur2VW0i+0FZoBeWmOWv5fVlFFaU0p2fjaWMAtlzjLKneUYGOwt2cvekr28m/EuAFGOKEYkjGBEwggSwxOJckQRGRJJtCOaSEckUY4ookKiSI5IVvDeibTkd0bQB+VHcrvdDTLhIp1JaZUZlEeH2RWQ+8nb2G3G0KbXazWrJBtWPGfePu3PrQrIoeFe5apsEDk2q9XCr2cN4abX1gBw2cT0AI9IoC4o79bd15tz8m+h5xhY8CfI22E2f1vxLMz5K/SbGejRSZALtYX6AuojOZ1O5s+fz1lnnYXD4cBtuCmqLmL94fWsyFnBxryNbCvYRpmzjBU5K1iRs+KYr9Uvph9DegxhVOIoRieOZmj8UEJsWg/Z2QU0KC8rKyMjI8P39Z49e1i3bh3x8fEkJCRw//33c95555GamkpeXh5PPvkkmZmZ/OAHPwjgqEVar7TK07wsXFfb/XXe2DSSo8MY1yeuZU9c9DC4qiB9Mgye3erX7xMfgdUCbgOiQjvddUyRgDhzZE9mDk3icGk1545NC/RwhHr7lHf37utNsVhgyBwYOBNWvwTf/N0Mzl//IbZ+pxATOifQI5QuwmqxEh8Wz8w+M5nZx7zg43K72FW0iy35W9hWsI2i6iLKneW+zHppTanvc3VtNdsLt7O9cDsf7f4IALvVzrAewxiZOJLRiaNJj04nOiSamJAYokOiCbeHKxHUCQT0L8xVq1Yxc2bdFchbb70VgGuuuYZnnnmGbdu28d///pe8vDwSEhKYOHEiixcvZuTIkYEaskiblHiC8ugwBXf+stusTB989H1LG8ndYf5hBebWN234xygy1M79F46moqaWWF1MEfGL1WrhxR9PCvQwpB67d59ytzLlzbI5zD3Lx/wQFv8Tlj2Nde8iZrCY2m29YfSFgR6hdEF2q52h8UMZGj/0qMfVumvJLMtkV9EuthZsZWPeRjblbaKouohN+ZvYlL+pwdZu9c/vDdB7RfViUNwgBvcYzOC4wQyIG0C4XdtVBoOARgYzZszgaH3m3nvvvQ4cjcjx5ytfD1Vwd1x9ea+5B+2QudBv+rGPP4bLJvVph0GJiASOQ/uU+y8sFmbfByf+BPf827Du/Azb53+EIbMhNCrQo5Nuyma10SemD31i+viy7IZhkFmWyaa8Tb4gPa8yj9KaUkpqSqg1anG5XRRUFVBQVcC+kn0szVpad06LjVGJo5iWNo1Te5/KiIQRyqoHiNJ1Ih2oxBOUx4Trf73jZt/3sO1jsFhh1j2BHo2ISFCwWb3l68qU+61HP2ovep7KRyYQWZYDSx6F0+4M9KhEfCwWC72je9M7ujdn9j+zwWOGYVDpqqSkpoTSmlKKqovYV7KPjKIMdhbuZGfhTgqrC1mfu571uet5ev3T9IrqxZTUKaREppAcnkxSRBLJEckkhSfRI6yHmswdRy2KDIqKinj//fdZvHgx+/bto6KigqSkJMaPH88ZZ5zBtGnTjtc4RbqEkkpv+boy5cfF/uUw7wbz9oSrIXlYYMcjIhIkfI3e3MqUt4g9jE29LmPynsdgyeMw/kro0S/QoxI5JovFQoQjgghHBD0jewIwsedE3+OGYZBdns33Wd+zJGsJ32V+R2ZZJu/ufLfJ89ktdgbGDeSMfmdwZv8zSY9WE8/25FdQnpWVxV133cVrr71GWloakyZNYty4cYSHh1NQUMDXX3/NP/7xD/r27cvdd9/NpZdeerzHLdIpecvXYxSUty9nJXz1V/j+ScCAmF4w445Aj0pEJGjYlSlvtZzYCbj7nYJ17yJYeBf88OVAD0mkzSwWC2lRaVw85GIuHnIxla5KlmQuYUfhDg5XHCavMo/DFYfJrcwlvzIfl+HyNZl7fO3jTE6dzKVDL+XU3qeq+3s78CsoHz9+PNdccw2rV69mxIjG7f4BKisrmTdvHo8++igHDhzgd7/7XbsOVKQrKFWjt/a3fzl8cBPke3ZyGHs5nPk3CO8R2HGJiAQRu7ZEaz2LhdrZf8X63Kmw5QPIWgdp4wI9KpF2FW4PZ1bfWczqO6vRYy63i7zKPL7P+p75e+azPHu57yPCHsGU1Cmc0vsUpveaTkpkSgBG3/n5FRls2bKFhISEox4THh7OZZddxmWXXUZ+fn67DE6kqympt0+5tNGR2fHoVDj3MRhyRqBHJiISdHyN3lS+3jrJI2D0D2Dj/8x/e658J9AjEukwdqudnpE9uXDwhVw4+EKyyrJ4Z8c7zMuYR25lLl8d+IqvDnwFwKC4QfSN6UvvqN6+9e69o3qTFpWmjPpR+BUZHCsgb+vxIt2F9ilvJ5lr4L2f1WXHx10BZ9yv7LiISDO8W6KpfL0NZvwRNr0LGQth/zLoMyXQIxIJiLSoNH454Zf8Yvwv2FqwlcUHF7P44GI25m0koyiDjKKMRs+xYKFvTF9+MOQHXDzkYiIdkQEYefBqc7pu69atLFu2jPHjxzNu3Lh2GJJI11W3plyZ8lYrzoRXL4bKAmXHRUT8pC3R2kHCQLPR25r/wpf3wTUfg1XdqKX7slqsjEwYyciEkdww9gbyK/PZnL+ZA6UHyCzL5GDpQQ6WHeRg6UEqXZXsLdnLw6se5pn1z3DViKu4euTVCs49WhQZ3HfffYSHh/P73/8egK+//pozzzyT6OhoiouLeemll7jiiiuOy0BFugJ1X2+jWhe8+1MzIE8dC1d/oOy4iIgf7L7ydWXK2+TUP8D6N2HfEnj3OrjwGbCHBnpUIkEhITyBU3qf0uh+wzDIr8rn6wNf8/Lml9lbspen1j/FG9ve4GdjfsalQy/t9qXtLbq898477zRo9Hb//ffzy1/+kry8PJ544gn+9re/tfsARboSdV9vo28fhP1LISQaLnlRAbmIiJ/qyteVKW+T2N5wwVNgdcDm9+C1H0BVSaBHJRLULBYLieGJ/GDID/jggg/4x6n/oF9MPwqrC3lo5UOc8/45vLfzPSpdlYEeasD4lSl/+eWXMQyDvXv3sm7dOvLz8zEMgyVLlnDyySfz8ssv43a72b17Ny+/bG4TcfXVVx/XgYt0RiXqvt56616HRQ+bt8991CwjFBERv9SVrytT3majL4GIeHjrKtjzLbx0FlzxLkSr67TIsVgtVs7odwan9zmdeRnzeHrd02SXZ3P30rv5x8p/cEb/MxiRMMJsFBfVm55RPXFYu34yy6/IoG/fvgCEhISQkpJC3759WbduHTExMcycORPDMKiursZisdCvXz8MQ7/wRY7kqnVTUVMLKChvsU3vwgc3AwZMucn8g0hERPzm3RLNqe7r7WPgaXDtx2amPGcjvDAbrngbkoYGemQinYLdaueSIZdwzoBzeHPbm7y5/U0yyzJ5Z0fDnQ2sFis9I3rSK7oXfaL7MDl1MtPSphEbGhugkR8ffkUGp556KgATJkzg448/5rbbbuOzzz7jrLPO4pRTzHUDGzduJD093fe1iDRUVu3y3daa8hbY+jG8+zMw3DDhGjhDy2RERFrKblWmvN2ljYfrFsArF0HhHnh6Gky63lx3ruVVIn4Js4dx7ahruXrk1azIWcHX+7/mQOkBDpYdJKssi+raarLKs8gqz2Jlzkre3fkuVouViT0n8uzsZ7FaukazxRal6x5++GHOP/98TjrpJIYNG8azzz7re+yll17izDPPbPcBinQVJZVmUB7msBJi7xq/QI67nV/A29eCUQtjfgTnPAoWS6BHJSLS6Tg8mXLtU97O4geYgfkHN8POBbDsKVj/Bpz6R5h4Hdh0EV7EH1aLlSmpU5iSWrfVoNtwk1eZ5+vkvqNwB99lfkdGUQZWrF0mIIcWBuVjx45l79695OfnN9qL/He/+x0xMTHtOjiRrsS7nlxN3o4trKYA6+r/wBd3gdsJIy6A85/U1jMiIq3k7b6ufcqPg6hks3Q940tYcCcc3gKf3QYrn4PZf4Ghc3VBWaQVrBYryRHJJEckMz55PAC/PfG3ZJVlUVpTGuDRta9WLWw9MiAHSE1NbfNgRLoyb+d1rSdvgqsa9i+DjC+wZ3zBGYe3wGbPY0PPgoufB5veNxGR1nJ4LmrWaku042fQ6TBgBqx9Bb76K+RnwJuXQb+T4bzHzay6iLRZWlRaoIfQ7vz6K/fNN9/kRz/6kV8nPHDgAPv37+ekk05q08BEupq6zuvKlANQsNvMKmR8CXsWgbMcAAtgYMFIm4B1+Nkw9Rcq/xMRaaO6TLnK148rqw1OuBZGXgRLHoWlT8DexWan9p8vMh8XETmCX7WgTz/9NMOHD+ehhx5i69atjR4vLi5m/vz5XH755UyYMIH8/Px2H6hIZ+fbozy8mweYzip4+Xx4fDzM/x3s+NQMyCOTYezluC54lk9HP0Htjz+Hk38L9tBAj1hEpNPzdl9Xo7cOEhYDp98Fv1gBYbFwaBOsey3QoxKRIOVXpvzbb7/lww8/5N///je33347kZGRpKSkEBYWRmFhITk5OSQmJnLttdeyadMmUlK0T6PIkUq1R7np89th9zdgtUP6FLPcb9AsSBkFViuG04lz3/xAj1JEpEtxeLuvq9Fbx+rRD069DT6/wyxpH3khhEYHelQiEmT8jg7OO+88zjvvPPLy8vjuu+/Yt28flZWVJCYmMn78eMaPH49VTZhEmuXtvh7TnYPyTe/Bqv8AFrMpzsDTAj0iEZFuwbdPuTLlHW/iz2Dl8+ayrSWPwWl3BnpEIhJkWhwdJCYmcsEFFxyHoYh0baXdvft6XgZ89Cvz9sm3KiAXEelA3jXlypQHgD0EZt0L/7sKlv4bxl4GCQMDPSoRCSJKbYt0kJLuXL5engevXQLVJdBnKsy4I9AjEhHpVrzd17WmPECGn2t2ZndVwYe/BF0cEZF6FJSLdJC6LdG6WabcWQlvXAaFeyCuL/zwZW1vJiLSwdR9PcAsFjj3MXBEwL7vYM1LgR6RiAQRBeUiHaSu+3o3Ckjdbnj/Bjjo6T57xTsQlRzoUYmIdDsOX/m6MuUB06Of2ZEdYMFdUJIV0OGISPBQUC7SQXzl66HdKFP+5T2wZR5YHfCj1yFpSKBHJCLSLdlVvh4cJl0PvU6EmlL46v5Aj0ZEgkSrg/Kamhq2b9+Oy+Vqz/GIdFndap/yqhJY9A+zyyzABU9Bv+mBHZOISDem8vUgYbXBmQ+Yt9e9Boc2B3Y8IhIUWhyUV1RUcN111xEREcHIkSPZv38/ALfccgsPPPBAuw9QpKvo8vuUV5fCxnfgjcvh4UHw1V/M+2f+Ccb8MLBjExHp5hyeLdFUvh4E0ifCiPMBA764J9CjEZEg0OKg/Pbbb2f9+vV88803hIWF+e6fNWsWb731VrsOrjtz1rr5wTNLueWNtYEeirQT7z7lXSooryk39x5/6yozEH/3Otj+CdRWQ8JgOONvcMrvAz1KEZFuz25VpjyonH43WO2wcwHsWRTo0YhIgLU4Opg3bx5vvfUWU6ZMwWKx+O4fOXIku3btatfBdWfbc0pZubcQKORfPxzru8ItnVOVs5Yazx9CXaJ8fecXsO5V2PE5OCvq7o8fACMvgpEXQspIs9usiIgEnC9TrjXlwSFhIJzwY1j5HHx6G/x8Edi6wN8HItIqLQ7Kc3NzSU5u3D25vLy8QZAubZNxuMx3u7C8huSYsKMcLcHOu57cYoGokE6eKV/zCnz4i7qv4/rCKE8g3nOMAnERkSBk93VfV6Y8aMy8Aza/B4e3wPdPwPTfBHpEIhIgLU6/nnjiiXzyySe+r72B+PPPP8/UqVPbb2TdXP2gPK+sJoAjkfbg7bweFWLHau3EQWvmGvjkt+btsZfDz76GX62HWfdA6lgF5CIiQcrbfd1Za2AYypYHhYh4mOPpwP7Ng1CwJ7DjEZGAaXHK7m9/+xtz585ly5YtuFwuHnvsMbZs2cLSpUv59ttvj8cYu6X6QXl+eXUAR9K53fzaGnbnlfP+TdMIc9gCNo7cUvNn2CMyJGBjaLPyfPjf1eZ68aFnwflPglXLKkREOgPvPuUAtW7DlzmXABv7I7ML+97F8N7P4IJnIHFQoEclIh2sxX9RT58+nXXr1uFyuRg9ejQLFiwgOTmZ77//nhNOOOF4jLFbysitC8oLypUpb43C8ho+2ZjN1uwSNmUWH/P4zKJKbv3fOr7bmdfuY9meUwrAkJSodj93h6guhdd/AMUHIH4gXPiMAnIRkU7EXq83jTqwBxGLBc55FBwRcHAlPDUZ5v/BvBAuIt1Gqxa3Dhw4kOeee669xyIezlo3e/PKfV+rfL11NtQLxHfnlnNiv/hmj80pruLy55axL7+Cg4WVTB+c2K5j2eYJyof2jG7X83YIZyW8cRlkrobwHnDZGxAWG+hRiYhIC9jrLZ1y1roDWj0mR0gcBNd/Awv+DDs/hxX/B+vfhFN+C5N+Dg71FRLp6lqc6po/fz6ff/55o/s///xzPv3003YZVHe3L7+8wVXsApWvt8r6A0W+27vqVR4cKa+smiueNwNygAMFFc0e21rbckoAGNozpt3PfVw5K+GtK82yupBouPI9SBoa6FGJiEgL1d/FRR3Yg1DSULjif3D1B9BzNFQXw8K74ImJsPEdUIM+kS6txUH5H//4R2praxvdbxgGf/zjH9tlUN1d/fXkAPnKlLfKhoNFvtu7csubPKawvIYrn1/OrtxykqJDAcgpqaLK2XiOt5bbbbDDkykf3pky5TXl8PqlkPEF2MPNPxZ6TQj0qEREpBVsVouvF6dTAV7wGjADrl8EFzwN0WlQvB/evQ7evEyBuUgX1uKgfOfOnYwYMaLR/cOGDSMjI6NdBtXdeYNym6fULF9rylvMMAzWHahfvt44U15c6eSq/yxnW04pydGhvP3zqUSE2DAMc315e8ksqqS8ppYQm5V+iZHtdt7jylkJr/0A9nwLIVFw5bvQd1qgRyUiIm3gsGqv8k7BaoVxl8Mtq+G0O8EeBjs+g1UvBHpkInKctDgoj42NZffu3Y3uz8jIIDKykwQcQW6nJygflWaWOueXqXy9pbKLq8ir977tL6jAWVt3hbms2sW1L65gU2YJCZEhvP6zyfRLjKRPfATQviXsW7PN0vWByVENygeDltsN710P+5ZAaAxc9T70OynQoxIRkTby7VWuoLxzCImAU34Ps/9ifr3wbijcG9Ahicjx0eJGb+effz6//vWvef/99xk4cCBgBuS//e1vOe+889p9gN2RN1M+eUAC6w8Wq/t6K3hL14enxrAvv5yKmlr25VcwKDmKyppafvLSStbuLyI23MEr101mULJZVt67RwTbckp9QfkH6zL591cZ1B7RqfbUIUn8+ZwRvmqGo/F2Xh/WWUrXv7wHtn4IthC4/C1InxToEYmISDvwNntT+XonM/GnsGWeebH8g1/A1R9qBxSRLqbF/0c/9NBDREZGMmzYMPr370///v0ZPnw4CQkJ/OMf/zgeY+xW3G7D15Rscn+zW7jWlLfc+oNm6fq49DgGJJkVHLtzy6hxubn+lVWs2FNAdKidV66bxIi0uuZrvkx5oVm+/n/f7ibjcBl78sobfLy0dC8PfLrVr7FsO9SJgvJVL8KSx8zb5z+pknURkS7EW62lTHknY7XC+U+Y/V32LoZ1rwZ6RCLSzlqcKY+NjWXp0qUsXLiQ9evXEx4ezpgxYzjllFOOx/i6ncyiSqqcbkJsVsalxwFQWu2i2lVLqF3bl/jL23l9bO9YyqtdbMosYXdeOcWVmSzemUdEiI2XfjKRMb3jGjwvPT4cgP35FVQ5a9nhCaifu/pEekQ4ANiSXcJdH2zmucV7GNYzhotP6H3UsWzL9nZe7wRBeblnj/YZd8CYHwZ2LCIi0q685ev1l3NJJxE/AE77Eyy409w6bchciEoK9KhEpJ20ap9yi8XCnDlzmDNnTnuPp9vbnGUGcP0TI4mPDMFuteByGxSU15AaGx7g0XUObrfBRk+mfEzvOA6VmGvLdx0u47viKgBuPHUgJ/RtvG95Xaa8gh2HSnG5DXpEOJg1PBmLp23tif3iyS2t5t9fZXD7+xsZmBzlu4BypCpnLXs9W60N6wzboZ36e+g3HfpMCfRIRESkndm9jd7cypR3SpNvhPVvwaGNZnB+0f8FekQi0k78Csoff/xxrr/+esLCwnj88cePeuwvf/nLdhlYd1Nc6eTJrzN4acleAEb2isFisRAfGcLh0mryyxSU+2tPfjml1S7CHFaGpET5lgOs3FvAfs9a8fPH9Wryud6gfH9BBRszzcB+VK9YX0Du9ZtZQ9iWU8rCLYe4/uVVfHTLdFJiwhqdL+NwGbVug7gIBykxoe32PR5XfacGegQiInIcOHyN3pQp75Rsdjj3MXj+dNjwptmhfcCpgR6ViLQDv4LyRx55hCuuuIKwsDAeeeSRZo+zWCwKylvBWevm7McXc9CzjvnkwYncduYwgLqgXM3e/LYv39yTfGBSFHab1bem3JuxntAnjj4JEU0+t3cP8/7SKhdLM/IBMyg/ktVq4ZFLx3HRU0vYcaiM619ZzVvXTyHM0XCJwTZPk7ehKdGNAnsREZGOZPesKXdqTXnn1fsEOPEn5vZoix5WUC7SRfgVlO/Zs6fJ29I+HDYrl56Yzgfrs/jT2cOZMSTJF8AlRoUCpdoWrQUKy52AeUEDYEBiVIPHm8uSA4SH2EiKDiW3tJqvth0GYFRa46AcICrUznNXn8j5Ty5h/YEi7nhvI//84Vjfz84wDOatzQRgZDPnEBER6Sje7usudV/v3E6+FVa/ZDZ9y9kIPUcHekQi0kYt6r7udDoZOHAgW7f613Va/Hf9qQP47FcnM3NocoOMakKUGVhqWzT/FVaY71WPCPO9Cw+x0SvOLP23WS2cPSb1qM9P72EeW+msBWB0E5lyr74JkTx5+QRsVgvvrc3k+cV1F60+35zDdxl5hNitXDutX6u/HxERkfag7utdRGxvGHG+eXvZ04Edi4i0ixYF5Q6Hg6qqquM1lm4t1G7zlZXV58325mlbNL95g3Lvewf4SthPHpzoqT5onnddOUBMmN3Xkb05Jw1K5M9nDwfg759u5eMNWZRXu/jLx+bFq5+fMqDZcnkREZGOou7rXcjUm83PG9+GssOBHYuItFmL9ym/+eabefDBB3G5XMdjPHIEbwBZUK7ydX8VVpjl63GeLcwAzhqdSqjdys9OHnDM56fXC8qbavLWlGum9ePSE9NxG/CL19cy/i8LySyqJC02jJtmDGrFdyEiItK+HOq+3nX0PhF6T4TaGlj5QqBHIyJt1OIt0VauXMmXX37JggULGD16NJGRkQ0ef++999ptcFKX7c1XptxvRUeUrwNcNqkPP5qY7leAfWRQ7g+LxcJ9F4zEZrPw8fosSqrMi1Z3njOC8BDtLy8iIoGnTHkXM+UmeOfHsPI5OOmXEBJ57OeISFBqcVAeFxfHxRdffDzGIk1I8Java02537zr7+tnygG/u5/XL18fmeb/3uKhdht/u3A09503knUHiqh2uTlpUKLfzxcRETmebN5Gb1pT3jUMPw969IPCvbD6vzD1pkCPSERaqcVB+Ysvvng8xiHNqGv0pvJ1fxVVNOy+3lL1M+VHa/LWHLvNyon94lv12iIiIseLr9Gbuq93DTY7TP8NfPQrWPo4TLwO7EfvmyMiwcnvNeVut5sHH3yQk046iYkTJ/LHP/6RysrK4zk2ARIizV+uKl/335Hd11sqNSaMSf3jmdw/nn4JKgUTEZGuwbslmvYp70LGXgYxvaA0G9a9FujRiEgr+R2U33///dxxxx1ERUXRq1cvHnvsMW6++ebjOTahLlNeUVNLZU1tgEcT/AzDaLLRW0tYrRbeun4Kb/18KlarfyXvIiIiwa5uSzRlyrsMeyic9Cvz9uJHoDw/sOMRkVbxOyh/+eWXeeqpp/j888+ZN28eH330Ea+99hpulUAdV1GhdkI8/4jmq4T9mCpqaqlxmXOytZly8H/9uYiISGfhbfSm7utdzISrISoFivfD01Nh5xeBHpGItJDfQfn+/fs566yzfF/PmjULi8VCVlbWcRmYmCwWS7115SphPxZv6XqI3UqEup6LiIj42D1boql8vYtxhMOV70HSMCg7BK9dDPN/DzUVgR6ZiPjJ76Dc5XIRFhbW4D6Hw4HT6Wz3QUlD3qA8r0yZ8mPxNnnrEeFQtltERKQehzdTrvL1rqfnKLj+G5h8g/n1imfh2RmQtS6AgxIRf/kdlBuGwbXXXstFF13k+6iqquKGG25ocF9LLFq0iHPPPZe0tDQsFgvz5s3zPeZ0Orntttt8e6GnpaVx9dVXd8vMfHK0eTHkUImC8mPxVhO0pXRdRESkK/LtU67y9a7JEQ5zH4Qr34WonpC3HZ6fBd89Am71JRIJZn4H5ddccw3JycnExsb6Pq688krS0tIa3NcS5eXljB07lieffLLRYxUVFaxZs4Y///nPrFmzhvfee4/t27dz3nnnteg1uoKUGDMozymuCvBIOt7h0irKq11+H+8tX29tkzcREZGuylu+rkx5FzdoFtz0PQw/F9xO+OIeePUiqFV1q0iw8nuf8uOxP/ncuXOZO3duk4/FxsaycOHCBvc98cQTTJo0if3799OnT58mn1ddXU11dV1GuaSkBDAz75211D4p0gwwc4orOt334B1va8ZdWFHDqf9YxICkSObdONWv5+SXmhcu4sK1tKKzasucke5Fc0VaqrvPGZvFzJBXO13d9j1oqU47ZxzRcOF/sAx8E9vnf8Sy+xtcG9/FGHlxoEfWrXTa+SPtoiU/d7+D8mBQXFyMxWIhLi6u2WP+/ve/c++99za6f8GCBURERBzH0R0/hw9ZABsbMw4wf/6+QA+nVbwXWKprodwF8aHHfs72YguVThubs0qZ99F8/OnbtuKAFbBSkpfN/PmZbRu0BNSRF+VEmqO5Ii3VXefMvn3mv5EZu/Ywf/6uQA+nU+m8cyaWIQlnMDznPUoXPMiivWGgnjsdrvPOH2mLigr/my12mqC8qqqK2267jcsuu4yYmJhmj7v99tu59dZbfV+XlJSQnp7OnDlzjvq8YBa5I5c3d6/FHRbLWWf5lzEOFk6nk4ULFzJ79mwcDgc3vLaWxRn5vPvzyQzrGX3U55asPAhbtgAwctIpDE6JOubrrfp4Kxw8wNihAzlr9uB2+R6kYx05Z0Sao7kiLdXd58yOLzP4Ims3vfv05ayzhgd6OJ1Cl5gz5ZMw/v0JPSp2c/aYRIz0yYEeUbfRJeaPtJq3YtsfnSIodzqd/PCHP8QwDJ5++umjHhsaGkpoaOM0rMPh6LT/M/SKN4PR3NLqTvs9eN//rdml1LjcfLghh9Hp8Ud9TmZx3TKErJIaRvQ+9vdeXGU2MkmIDuu075WYOvP/s9KxNFekpbrrnAl1mH/2uaFbfv9t0annTFwqjL0U1ryMfeUzMGB6oEfU7XTq+SOt1pKfud+N3gLFG5Dv27ePhQsXdtpsd1t4G73ll9dQ7Wq6e+aBggr+8M56fvnGWn715lq+3na4I4fot5Iqs2nbJxuycR+j++v+gnLf7X0F/pV/eBu9qfu6iIhIQ77u69qnvPuZcpP5edsnULg3oEMRkcaCOij3BuQ7d+7kiy++ICEhIdBDCogeEQ5C7OaP6nAz26Ld+9Fm/rfqIB+uz+KDdVnc89HmjhyiX2rdBmWeTupZxVWsPVB01OP35dcF4vvzy49yZB1fUB6pq5EiIiL1OdR9vftKHg4DTwfDDSueC/RoRFott7Sad1cfZNnu/EAPpV0FNCgvKytj3bp1rFu3DoA9e/awbt069u/fj9Pp5JJLLmHVqlW89tpr1NbWkpOTQ05ODjU1NYEcdoezWCykxJgl+YdLG2+LtiWrhC+2HsZigetPGQBAQVnwvUdlVQ23NvtkQ3azxxqGwf56QbnfmfJys8thnDLlIiIiDWif8m5u8s/Nz+teA2dlYMci0kpbs0v47dvruefD4EtAtkVAg/JVq1Yxfvx4xo8fD8Ctt97K+PHjueuuu8jMzOTDDz/k4MGDjBs3jtTUVN/H0qVLAznsgEiJ9u5V3jhT/uQ3GQCcMybNF5SXVruoDbJ/dEuqGm4LMH9j8yXshRVOSuvtT14/QD+aIk+mPF5BuYiISAN2mzLl3dqgWRDbByoLYfP7gR6NSKvklpqxUFK0H1s5dSIBbfQ2Y8YMDKP5wPFoj3U3KbGeoLykYaY843AZ8zeaGeebZw4kOqzuR1pW7SI2PHjKuL1BeVyEg9pag5ySKtbsL+TEfo0bvu3zlKuH2KzU1Lo5UFhBrdvAZm1+G49qVy3lNeaae60pFxERacjh+TfUpTXl3ZPVBif+GL68F1Y+D+MuD/SIRFosr8wMyhOjulZQHtRryqVOT0+zt0NHBOVPf7MLw4DZI1IY1jOGULuNUM/685JK/zes7wgllWbmOyEyhNkjUgD4uJkS9v2ecvWx6bE4bBactQbZxUcvtSqqML9fq4UGFydERESkLlOu8vVubPxVYAuBzNWQuSbQoxFpsbqgvGsl4BSUdxLeNeX1g/IDBRXMW5cJwC9mDvLdH+PJjpcesYY70Eo9mfKYcAdnj0kFmi9h9zZ565cQSXqPCODYJezeJm9xESFYj5JRFxER6Y4cNm+mXOXr3VZUEoy4wLy96oWADkWkNfI8fbOUKZeA8G6LllNcF5Q//e0uat0GJw9OZGx6nO9+b5b4yDXcgebdDi06zMH0wYlEh9k5XFrNyr0FjY71BuV94iNIjzeD8mM1eyso926HFjwl+yIiIsHC7uu+rkx5tzbxp+bnje+Y68tFOhHvmnIF5RIQR5av5xRX8c6qgwDcctrgBsfGhJlBafCVr3sy5WF2Qu02zhjZE4BPNjYuYT/gCcD7JETQN8GTKT9GUO4tX9d6chERkcbquq8rU96tpU+ClNHgqoJ1rwd6NCIt4i1f72qN3hSUdxIpvqC8GsMweHbRbmpq3UzqF8+k/g0bpQVv+bo5Hu/46krYcxp1it9XYDZ665sQSZ/4uvL14kon//luD59uzGZ/fkWDZoD1y9dFRESkobrydWXKuzWLBSb+xLy98gXQRRrpRLpqozd1w+okenq6r1c6a9mbX8HrK/YB8IvTBjU6NnjL183xeMd30sBEYsMd5JVVs2JPAVMHJgBQ5azlUIn5P1zf+Ahfmcre/HJueWMti3bk+s4ZHWpneGoMI9JiOFhoNoJT+bqIiEhj3vJ1p9aUy+gfwoK7oGAX7PkGBp4W6BFJJ2UYBre/t5HYCAe3zx1+XF+r1m34lqsmRnetJJwy5Z1EmMPm297s/k+2UuV0M6Z3LCcPTmx0rLd8Pfgy5d7ydXN8IXYrZ4w0u7B/sjHLd5y3TD06zE5chMNXvr45q4RFO3IJtVsZ1SuGEJuV0moXK/YW8NLSvXyx9RAAPSK71v+kIiIi7cFbvu5S93UJjYJxl5m3V6rhm7Reblk1b648wP99u/u4JwTzy6txG2axR3wXq4xVprwT6RkTRnGl0xd8/mLmICyWxl3GY8I9mfKgW1PuKV+vt13Z2WPS+N+qg3y2KYd7zh2J3Wb1NXnrmxCBxWLxla97/ens4Vw9tR/OWje7csvYklVifmSXUFBew7lj0jrumxIREekkHDZvozdlygU48TpY8Sxsnw8lWRCjv5+k5cqra323d+eWM65e8+n2lldqZskTIkN8Wzx2FQrKO5HkmFC2HyoFYGhKNLOGpzR5nK/RW5CWr3vXlANMG5hAjwgHeWU1rNhTwLRBiWQcLgOgb3wkYFYJpMSEcqikmplDk7hqSl/A/ONiWM8YhvWM4aIJHfzNiIiIdDJ2z3ahTq0pF4DkYdBnKuz/Hta9Bqf8PtAjkk6ovLquMnd3btnxDcq76HpyUPl6p+LtwA5w82mDmt2L25uJDr7ydW+mvC4od9isnDnK7ML+8cZsDMPgA8/e6/Ub2N1w6kBOHZLEQ5eMbbI6QERERI7OlylXYy/xGn+V+Xntq2r4Jq1S6azLlO/KLTuur6WgXIJCqqfZW//ESM4endrscd5MdLBmyqPDGhZonD3aLJf6bFMOq/cVsi2nlFC7lQvG9fId8+OT+vPfn0zqctsfiIiIdBS7uq/LkUZeACHRULgX9n0X6NFIJ9QwU15+XF+rbo/yrrWeHBSUdyoXTujNyYMTeeCi0diayZJDXdAbtJny8Ibd0acMiCc+MoSC8hpue3cDAGeNTiVWXdRFRETajbf7uhq9iU9IJIy+2Ly95pXAjkU6pcqajs+Ud8UknYLyTqR/YiSvXDeZyQMSjnqcb015EDV6MwzDN5765esA9nol7Ls8V9h+NDG9YwcoIiLSxdXtU64yZaln/NXm560fQmVRQIcinU9FvaB8b14Ftcfxol9emWc7NJWvS2cQ7Wv0FjyZ8kpnre/K/JHl6wDn1CvHH5AY2WA9uYiIiLSdt1uxU5lyqa/XBEgeAa4qeOtKqCwM9IikE6moqYs3amrdHCysOG6vpTXl0ql4t0QrrXJiGMHxD6+3dN1mtRARYmv0+KT+8b71IZdOTFczNxERkXbmsCpTLk2wWGDuQxASBXsXw/OzIH9XoEclnUT9TDkc33XlvjXlKl+XzsBbHu6sNahyBsc/vN6sfXSYvcmA226z8tcLRvHDE3tzhWfLMxEREWk/3ky52wC3suVSX/+T4SefQ0xvyM+A50+HvUsCPSrpBMqPCMqP57py35pyZcqlM4gIsfkawQVLB/amtkM70pmjUnnokrFEhTYubxcREZG28XZfB3C2cvurL7ce4m/ztyrb3hX1HAU/+wrSJpgl7C+fD+veCPSoJMhV1jRcLrvrOGXKa90GBeWeNeXR6r4unYDFYqnXgT1YgnJPk7dwBdwiIiKBEGa34d28pbC85X8f1LoN/vDOBp5dtJtvtue28+gkKESnwLWfwIjzwe2EeTfAl3/RHubSLG+mvH9iJAC7j1OmvKC8BrdhrraIj1BQLp2ENyNdXBkczd585euh2uZMREQkEELsVgYmRQGwJbu4xc9ff7CIfE+manNWSbuOTYJISARc8hKc/Fvz68X/gHd+DLXBkeiR4OLdEm1Ur1jg+GXKvevJ4yNCfEtxupKu9x0JQBBmyr17lCtTLiIiEijeP5w3ZbY8qP5q62Hf7a3ZCsq7NKsVTr8Lzn8KrA7YMg9WvxToUUkQKq82/8YflRYDmOu+i4/DtsxdeY9yUFDeZcUE2bZopb5Gb8qUi4iIBIo3KN+Y2fJM+Vfb6oLyLQrKu4fxV8Ccv5q3lz2tMnZppNJpZsqTY0JJ9gTMx6OEvStvhwYKyrssb0a65DhcqWoNb8O5ozV6ExERkePLm83a3MKgPLu4ki3ZJXg3UNlfUBE0zWTlOBt/JYTGQsEuyPgi0KORIOPNlEeE2BnaMxqAFXsK2v116oLyrreeHBSUd1nejHRpkGXKVb4uIiISOCM9mfKs4iryPX/kHqnKWcs32w9z70ebueipJby35iBfbzMbu41LjyMtNgyAbdmlHTNoCazQKJhwlXl72VOBHYsEHe8+5REhNuaMSAFg/qacNp/XMAwOlVTx9bbDPPHVTt5bkwl03Uy5IqQuqq58PTiuYpeofF1ERCTgokLtDEiMZHdeOZuySjh1SBKGYbA7r5xvt+fy7Y5clu3Op9pVV6a8Zn8RCZFmdur0YcmsO1BEVnEVW7NLmNQ/PlDfinSkSdebAfnur+HwVkgeHugRSZCoC8rtnDGqJ3d9uJn1B4o4WFhB7x4RLTpXQXkNL3y3mw0Hi9mSVeJrLFmfNxvf1Sgo76KCrdFbmW+fck05ERGRQBrZK9YMyjOL6RUXzs9fWdWoY3JqbBinDkkixG7l5e/3+f44njksmRqXmy+2HmZLVgll1S5ufm0NeWXVJEWHkhQVan72fkSFMq5PHKF2GwBLM/J49Iud3HXuCN/6dukEevSFYWfD1o9g+TNw7mOBHpEEifqZ8uToMCb1i2f5ngI+3ZjDz04Z0KJz/ee7PTz59S7f11YLDEqOYkRqDCPSYhjbO67LXghUhNRFxYR7MuVBsyWad59yZcpFREQCaXSvGD5an8WmzGK+2X6YXbnlOGwWJvWP59QhScwYmszg5CgsngXkY3vHcft7G+mbEMGI1Bj251cAZrO3F7/bw7c7jr5n+alDkvjvTyYB8N/v97JibwG/e3s9H98yvUtubdRlTbnJDMrXvwmn3w0RXTM4kpaprPGuKTcvvJ09JpXlewr4ZGN2i4Ny77rxs8ekcv3JAxjaM5owh619BxykFJR3Ud6MdPCVr2vKiYiIBNKoNDNDvWDLIWrdBuEOGwtvPaXZUtOLT+jNyUMSCXfYsFgsjPA0i9t+qJR9+WaG/dezBpMWG05uWTW5peZHVnEla/cXsWJPAW63gdVqIeOw2ZV5W04pr6/Yz9VT+x3/b1jaR5+pkDoWstfD6hfr9jGXbsswDCqcdeXrAGeO6sndH25m3YEiMosq6RUX7vf5Sj1N407s24Ox6XHtPt5gpsuTXVT9Rm+5pdX8+8udFFU0XpfRUerK15UpFxERCaSRnqC81m0A8IvTBh1z7WdydJjvb4v0HhFEhdqpcbkpqXIxODmKW04bzA8npnPzzEHcc95InrxiAm//fCohNiuVzloOFlZS43Kzz5NlB/jH59ubbTYnQchigck3mrdXPA+1wZH4kcCpcroxzF8jvkx5cnQYE/uZVRSfbsxu0fm8ndwjQ7tfEk9BeRdVf0u0P8/bxD8X7uCV7/cFbDzaEk1ERCQ4xEY46BNvBuF9EyL46cn9W/R8q9XCsHrNln41azA2q6XRcXablYHJUUBdVt3lNogMsTE8NYaSKhePfrGz2depcbn52curuOGV1bg9FxAkwEZdBJHJUJoFWz4I9GgkwMpr6pbJhtcrMz97dCoAn7QyKI9SUC5dhTf4zS6u4outhwA4WFgZkLHUuqHSaXZx1ZZoIiIigTdnRAohNit/OX+UrwlbS3hL2IemRHPWqNRmjxuSYgblOw6V+krXByVHcduZQwH4fHMOhtF0wP3k1xks3HKIzzbnsCu3rMVjlOPAHgoTf2reXvZ0YMciAVfpafIW7rBhrXdhbu6onlgssHZ/EVlF/scfZdXm+ZQply7DG5SXVbtwea4u5waoRKyytu52d7zyJSIiEmz+dPZw1tw1m1OGJLXq+VdP7cdpw5J56JIxDf4YP9KQFDOjvrNeUD4wOYopAxKwWy0cLq0ms4k/2jdnFfPk1xm+r9fsL2zVOOU4OPEnYHVA5irIWhfo0UgAeTPlkaENL+wlx4Qxsa9Zwj6/Bdnyukx592juVp+C8i6qqYz04dKqAIwEKjyVLZEhNnVZFRERCQIWi6VNF8oHJUfxn2snHrMZ0+Bkb6a8jIzcukx5mMPmy7av2V/U4DnOWje/e3sDLrdBiN38u2HNvobHSABFJcGI88zbq18M7FgkoLzboYWHNA6izxrdE2hdUK5MuXQZTf1Dm1samEx5QbV5BT21Bd0XRUREpPPzZsp35ZaxPacUgMHJ5n0T+vQAYO0RWfCnvt7F1uwSekQ4uPe8keYxB5QpDyon/Nj8vPEdqC4N7FgkYCq85eYhjeOOuaNTsVjMi27+lrCXeYPyJs7X1Sko76LsNiuRnqtWpw9LBiCvrMbXabUj5XuuBXibyoiIiEj3kB4fQajdSrXLzTZPUD7Ikz0f3ycOaJgp35JVwr+/Mpu/3Xf+KGaPSAFg5+GyoNnmVYB+0yFhMNSUwca3Az0aCZAKT/l6U5nylJgwTuxrXnj7dFPOMc/lrHVT7TJ7UHXH5a4Kyruw/kmR2KwWbjl9MBaLufVJYQC2RcurMjPlCspFRES6F5vVwmBPszeAEJuV9B5m5Zw3U74lq5gqZy3OWje/f2c9LrfBGSNTOGdMKolRofSJj8AwYN0RZe4SQBYLnHCteXvVi9BMsz7p2rzl681lts/ydGH3p4TdW7oOKl+XLub/rjqReTedxLj0OOIjQgA4XNLxJex5nqXsfRMUlIuIiHQ3Q5Lrtk/rnxjp6y/Tu0c4iVGhOGsNNmcV88w3u9icVUJchIO/XDAKi8W8qO/NqK89TkH55qxibnljra8Rnfhp3OVgC4WcDZC1JtCjkQA42ppygLmenRlW7ysku/joJeze0vUQu9XXS6I76X7fcTfSKy6c0b1jAUiKDgUC04E935MpV1AuIiLS/QxOqQvKB9XLmlssFl/A/fryAzzuKVu/97yRJEeH+Y7zZtSPRwf2kion17+8mo/WZ/HPBdvb/fxdWkQ8jLzAvL3qPwEdigSGt3w9spmgvGdsvRL2jUcvYS/3rE/vjqXroKC82/AG5YdLmu/AXuWsxd3Oa84Nw/BlyvvER7bruUVERCT4DakXiA9KimrwmDfgfnfNQZy1BrNHpHDe2LQmj1m7v7Dd/065+4PNvi3Zvtx6mOIKrVtvEW/Dt03vQVVxYMciHa4uU958IO1vCbuvyVs33A4NFJR3G94rzs1lytcfKGLcfQv443sb2vV1C8prqHZbsFggPV7d10VERLqbIfUz5clHBuVxvtux4Q7ur1e27jUsNZowh5WSKpdvW7X28OH6LN5fm4nVAikxodTUuvl4Y1a7nb9b6DMFkoaDswI2/C/Qo5EO5g3KI5rJlAPM9WyNtmpfITnFzScHy7tx53VQUN5t1GXKmw7KH/p8G1VON2+vPsi+/PJ2e939BebV554xYYTau+eVLxERke6sV1w40Z6S1KE9oxs8NqZ3HCGeNeb3nDeC5JiwRs932KyM8+yHfvlzy3h9+X5cte42j+vhz7cBcMtpg/np9AEAvLcms83n7VYsFjjRky1Xw7du51jl6wCpseGc4OvC3ny23BuUq3xdurTko6wpX747nyUZ+YD5u/TFJXvb7XX3F5pBubfTqoiIiHQvVquFf106jj+fM6JB1hzMBlGPXzaOv14wigvG9Wr2HHedM5IBiZHkldVwx/sbOfvx71i8M7fVYzpUUsWBgkqsFvjZKQM4f1waVovZkGpvXvslJ7qFMZeCPRwOb4aDKwM9GulA/pSvg38l7HXl6wrKpQvzNXprIlP+yBc7ABiZFgPA26sOtNteoPsLKgBthyYiItKdzR6RwnXT+zf52JmjUrlySt9GZev1jUiL4bNfn8Jd54wgNtzB9kOlXPXCCm5+fQ3OVmTN1+wzm8YN7RlDVKid5Jgwpg9OAuC9tcqWt0h4HIy6yLw97yYo2h/Q4UjH8WXKj7EO/Kx6JeyHmulv5cuUhykoly6suUz50l15LNtdQIjNyrNXn8jQlGjKa2p5a8WBdnndA76gXJlyERERab0Qu5WfTO/Pt7+fwY9P6ofdauGTDdnc9u4GjBaWTa89UAQ0XNN+8QQzU//+2oMtPl+3d+ptENMb8nfCC2fAoS2BHlGXUuWs5aUle47asDkQfJlyx9GD8tTYcIanxmAYsOFg0w0BvZnyKK0pl66sqe7rhmHw6EJz+5FLJ6bTKy6cn0zvB8BLS/dS2w4dTr1rypUpFxERkfYQFxHC3eeO5LmrT8RmtfDemkz+sWB7iwJpb6Z8vKezO8CcET2JDLFxoKCSVfvaf/u1Lq1HX7huASQNg9IsePFM2Pd9oEfVZTy/eDf3fLSFfy3cEeihNFDh2cbMn5LzXnFmgi6nmQsLZS04V1ekoLyb8DZOKa+p9ZWHLMnIZ8XeAkLsVm6aORCA88f1IirUTmZRJdtzStv8uvuVKRcREZHjYOawZO6/YBQAT369i9P/+S1Pfp1BdnHlUZ9X43KzIdPM1tXPlIeH2HxrX99bc/D4DLoNlu/O54Xv9gRvFj+2F/z4U0ifbG6P9soFsP3TQI+qS1i8Mw+ALdklAR5JQxVOM6YIP0qjN6+esUffnrmu0Vv3bAytoLybiAyx+UpLckurMQzDt5b88kl9SI01g+Ywh43RvWIB2HCwyO/z17jcXPXCcq57aSUF5TWAuc4kt8y8rUy5iIiItLcfTerDn84aTrjDxu68ch7+fDsnPfAVV/9nBR+tz6LKWdvoOVuzS6hxuYmLcNA/MbLBYxd6Stg/3pDd5HMD6U/zNvGXj7ewYk9BoIfSvIh4uGoeDDkTXFXw5hWw5pVAj6pTq3LWsnZ/EQC7Dpe1+aJMcYXzmBeu/OXLlPtRct7TkyBsblu0cjV6k+7AYrGQHFO3rnzRzjxW7ysk1G7lxhkDGxw7Jt0Mytc3s+ajKZ9vzmHxzjy+3HaYC59awq7cMg54StfDbQax4Y52+k5ERERE6vzslAGsvHMWD10yhkn94nEbsGhHLre8sZZJ93/BPR9u9q1XBViz31O6nh7XqLnclP4JpMWGUVrl4suthzv0+ziWPE9foObW5AaNkAi49FUYdwUYtfDhL2DxP7VdWiut2VdIjaeZYXlNLYea2d7YX1f9Zzlz/rWoXdan+7NPuZe3arf58nUF5dJNJEXV7VX+iGdNyhWT+5JyxJ6g43rHAbDe0wRlb145cx75lteXN99N89Vl+wCwWy3sy6/gvH9/xx/f2wBAYuMtR0VERETaTVSonR+emM7/bpjKN7+bwS2nDSItNoySKhcvLd3LJU8vJavITBas8WQdJ9RbT+5ltVp82fLjVcL+7KJdzPzHN2QW+Z+tdLsNSirNnXE2ZgZ5UA5gc8D5T8JJvza//vI+WP1SIEfUaX2/O7/B17tyy1p9Lmetmw0HiymtdrHIUxLfFuWe7uv+BOXeTPnhZi4qeM+lfcqly/Nmyv+36gDrDhQR5rByw4wBjY4bkx4HwPZDpVQ5a3ll2T52HCrjjRVNB+UZh0tZvqcAm9XCvJtP4sS+PSivqSu1SQzTlVERERHpGP0SI/ntnKF8d9tpvHjtRBKjQtmWU8r5Ty7hxSV7WLXXLP8e30RQDnDh+N4AfLMjl/yytmUlm/Ly9/vYk1fOsl35xz7Yo7zGhbf/7qasThCUA1gsMPteszM7wLKnlC1vhWWeoNxuNas6Mg63PijPqzefv2/B/GuKYRhU+jLlfpSvxx4rU65Gb9JNeDPl3+7IBeCqKX1Jjm6cxk6LDSMxKoRat8HmrGI+25QDwO7cptexvLrMDNZPH5bMqF6x/O/nU3n3xqn88rRBzB6ezMy0lu8fKiIiItIWVquFmcOSmXfzNIb1jCa3tJp7P9pCdnEVFguM9SzXO9Kg5CjG9o6l1m3w0cacBo+53QbXvriCMx5Z1Oza2KPJL6vmYKGZIff24PFHSVVd+f2evPIG5fhBb9ot4IiEvB1wYHmgR9OpVNbUss5TuTpnZArQtkx5/Sz1st35vr/r1x0oatF8BKipdePyXCmK8KM5W4on5iiudDbZr6Gu0ZuCcunikuuVqYc7bPz81IFNHmexWBjjKWF/bfl+X3lVU+tYKmpcvOsp77pySl/A/EfwhL7x3DpnKE9dPo6+Ue39nYiIiIj4p3ePCN65cRp/Oms40wYm4LBZOGNET6LDmu93c9EEM1s+b11Wg/s/25zDN9tz2X6olGtfXEFJlbNFY9lQr/Q8r9z/LLy3dB3MZPOWrODqwn1UodEw8kLz9pqXAzuWTmbVvgKctQapsWHMHJoMtDEoL62bc5lFlRwsrOTrbYe54Mkl3Pjq6hady5slB4g4xj7lADHhdsIcZuh5qIlseVmVgnLpJryZcoCrp/Ulsd7XRxrrCcrfX5vZ4P7dR/wi+Hh9NqVVLvomRDB9UGL7DVZERESknUSF2vnZKQN4/WdT2PaXuTxz1QlHPf7csWnYrRY2Z5WSbe7uSq3b8PXksVhgW04pN766mhqX/xWB3n49APllLciUVzYM/jd1hnXl9U242vy8+X2o6kQXFALMW7o+dUACg5LNLFdbytcPlzYMhr/flc/T3+4CYPmeAvbll/t9rnJPUB5is2K3HTuktFgsR+3AXtd9XVuiSReXGmf+jxAZYuPnpzSdJffydmD3Vqt7r2wdeXXu1eVmg7fLJ/XBam3YwVREREQk2Nj8+HslPjKEmcPMzOTKXPNvoE82ZrPzcBnRYXZe++lkIkNsLMnI57Z3N/i9TVX9zuktKRcuPjIo7yzryr3SJ0HiUHBWwKZ3Az2aTsO77nvKwAQGJJlB+aGSakpbWKHhdWTF60tL9zbYYu+DIypDjqbS2+StBUG0t2r3UGnDcRiGoUZvgR6AdJxpAxO5/pQBPH7ZeOIjQ456rDdTDuYVMG/Tk125dVfQNhwsYsPBYkJsVi45ofdxGbOIiIhIIFw03uzCvirPQmZRJY99YWbJf3byAKYNTOSpK0/AZrXw/tpMHv58+zHPZxgGGw4W+b5uSRM575pyh828oNDpMuUWC0y4yrytEna/lFe7fBdxpg5IIDbcQVK0WeW6O9f/jHZ9uZ5M+aR+8QBsyTarFrzVs/PWZfp9gcm3HZofpete3kz5oSMy5ZXOWl8jQzV6C4BFixZx7rnnkpaWhsViYd68eQ0ef++995gzZw4JCQlYLBbWrVsXkHF2FTarhTvOGs7pw1OOeWx8ZAjp8eEATB+cyNjeZuZ8d17dL4HXPA3ezhrdk4SjlMKLiIiIdDanDU8mJsxOcY2FGf9czK7ccmLDHfz4pH4AnDokiQcuGg3AU9/s4uXv9x71fFnFVeTVK1nPb0mjN0+m3NsxPuNwWYM1vZ3CmB+B1QFZayBnY6BHE/RW7i3A5TboFRdOenwEAIM82fLWriv3Nno7c1RP3wUegH9fNp5Qu5XdueVsPka/ggWbc7jx1dVszykFIKIFQXRzHdi9jQstFv+2V+uKAhqUl5eXM3bsWJ588slmH58+fToPPvhgB49MwPzHBuDiCb0Z6FnHssuzjqW40skH68315t4GbyIiIiJdRajdxsWePcutFpjQJ46nr5jQoEHcD05M59bZQwC464PN/G/VgWbP511PHhtuPr9Fa8o95cqDkqNIjArFbcDWnE62NjsqCYadZd5e80pgx9IJLNttlpVPHZjgu29gciTQ+nXl3kZvfeIjGJ9uXuCZ1C+eqQMTmOVJ2s07op9UfXll1fz27fV8uimHO+dtAloWRCd7Mv1HNnor926HFmLHYumey2EDWh8wd+5c5s6d2+zjV11llrns3bu3g0Yk9f3prBH8aGIfRvWK9ZVYZRVXUllTy/trDlLldDM0JZoT+ja9z6eIiIhIZ/bb2YOJKt7F1efNJik2osljbjltEIUVNby4ZC+3vbuBULuV88f1anTcek/p+qlDkvhwfRaVzloqalx+7fFcUmlmEmPDHYzuFcPX23PZlFnMhGb2Wg9aE66GLR/Ahrdg9n3gaLw1r5i+r9fkzWtgWzPlnvL15JhQrj2pHzklVdw2dygA549L45ON2by75iA9Y8OYMTSJgUlRDYLkhz/bTqlnKUW1p8FhS4Jyb6a8cVDevZu8QYCD8uOhurqa6uq6NTolJeZVRKfTidPZuqYI3ZXdAkOTI3A6nUSHWIgNt1Nc6WJnTjGvLDMbvF02sRcuV/N7ZXrfc7334i/NGfGX5oq0lOaMtJTVqGVQDEQ6jj5vbj9jMFU1Lt5YeZBb/7ceGwZnjGy4XHD9/kIAJveL47PNOdS43BwqqqB3j/BjjqOowvzbNtJhZXByJF9vz2XnodLON5fTp2OP6YWlJBPX5nkYIy8O9IiOq9b+zimtcvn6BpzYJ8b3/H6epaUZh8tafM5at+FbPtEj3MbwlERmD5vuG99JA3qQFBVCblkNf/1kK3/9ZCu94sI4eXAipwxKJDLUxlueSpC/XTCCBz7bQUmVizC71e+xJESYoWd2cVWD5xSVm0F6ZIi9883po2jJ99LlgvK///3v3HvvvY3uX7BgARERTV/hFP/0sNkoxsLD7y5hV66VEKtB+KFNzJ+/6ZjPXbhwYQeMULoSzRnxl+aKtJTmjLSUP3Nmkg12JVlZkWvlV2+t47qhbkb2MLtX1bph3X4bYKF4zwYirDZqsPDRgq/pG33s19+51wpY2b9rG043gI1NO/cyf/7utnxbATE0YiLDSjIp/OIxlu479gWJrqClv3M2F1qoddtICDVYt/Rr1nnuL6wGsLMnr4yPPp6PHzuR+ZTUQK3bjgWDlYu/wtZElfjNQ2BdvoWtRRYySixkFlXx5sqDvLnyoO+YiUluIg9t4NqB8M4eG+nuQ8yfP9+vMeRXmePPKargk0/m403CbyqwADZclWV+n6szqKio8PvYLheU33777dx6662+r0tKSkhPT2fOnDnExMQEcGSd37dVm9i7NouluTbA4MIJ6Vx03oijPsfpdLJw4UJmz56Nw+E46rEioDkj/tNckZbSnJGWaumcmes2+O07G/lkYw4vZTh49srxnDQwgX8u3ElV7R7iIx38+KJZfPLscoqyShk2biIzhyYd87yvZa+EwkJOOnE8bsPg/b0bsUcncNZZE9vj2+xYxaMxnviApLItnDVpMCQODvSIjpvW/s7Z8Nl2YB+njerNWWeN9N1vGAYPbfqKippaRk4+lQFJkX6fc3NWCaxeRkJUKOeePaPZ467wfK6ocbFibyGLduSxaGc++woqiAt38OiPp/nWht/i96ubql1u7lv7BS7DwrSZs+gRYe4G5VqfDds30islgbPOOrGFZw1e3optf3S5oDw0NJTQ0MadwB0Oh/4BbqPBKTFAFrWePQuumtrP7/dU77+0lOaM+EtzRVpKc0Zayt854wAe/dF4XO41fL75EDe8tpZfzxrC/y3eA8D9F4wmPCyUxKgwoJSiqlq/zlvqaYTVIyoMhyc9ml9e0znnceIAGDoXts/H8f2jcNGzgR7RcdfS3zkr9hYBcNKgpEbPG5gUxcbMYvYVVjE0Lc7vcxZWmnMoOTrMr7HEOhzMHhnO7JFpABwoqCA8xObbPq01HA7oEeGgsMJJQWUtybHmOKpqzdgiKqxr/W5uyfeifcrFb/Wvxo1Lj2NUr9gAjkZEREQk+DhsVh6/bDwzhiZR5XTzwKfbMAz4wQm9mTs6FYCEKDNDWODntmjeLdFiwh0kRZvPzSv1f5/zoHPqH8zPG9+GvIzAjiXIFFc62ZxlriefUq/Jm9dAz9/ju1q4V3n9Jm+tkR4f0aaA3CvFs1d5Tr29yr2N3qK66R7lEOCgvKysjHXr1vn2H9+zZw/r1q1j/35z/+uCggLWrVvHli1bANi+fTvr1q0jJycnUEPu1gbWC8q1DZqIiIhI00LtNp658gSmebaz6psQwd3n1ZUhJ0SagbV3d5tj8W6JFhNm9wVGJVUuql2dbK9yr7TxMGQuGG5Y9HCgRxNUVuwpwG1A/8RIX7fy+rwd2Fu6LZp3j3Jv6XmgNNWBvcy7JVo37r4e0KB81apVjB8/nvHjxwNw6623Mn78eO666y4APvzwQ8aPH8/ZZ58NwI9+9CPGjx/PM888E7Axd2d9EyLplxBBv4QIzhmTGujhiIiIiAStMIeNF66ZyIMXj+bN66c0yAImeAJrf/Yqr3Ubvm2oYsIdxIY7cHi6dOW1YK/zoDPjNvPzxv9B3s7AjiWILPNshdZUlhxgYHLrtkXz7lHuzVQHSkq0NyivuyBVtyVa982UB/Q7nzFjBoZhNPv4tddey7XXXttxA5KjctisfP6bU3C7zX9oRERERKR54SE2Lp3Yp9H98d5MuR/l62VVdVvPxoQ5sFgsJEaFkl1cRV5pNb3iOmkH87TxMORM2PEZvH4pXPsxxKQFelQB9/0uz/7kA5sOygfVC8q9cdTmrBKG9oz29RtoijczHehMeZpnvu7Lr+tM7g3Ko7txUK415dIioXYb4SEKyEVERERaKzHKG5Qfu3zdW7oe7rARYrd6nm8GVnl+lr8HrbP+AXF9oGAXvHQ2FO0P9IgCqqiihq05ZsfuKf3jmzymb0IEVou5l3luaTUvLtnLOf/+juc9zQSb482UJ0UHNlM+ure5G9baA4W++8qUKVdQLiIiIiLSkRIizaC6wI/y82Jfk7e6gMUb1Hf6oDwuHa79xBOY74bHxsErF8LaV6GyKNCj63DLdhdgGGYfp+RmysxD7Tb6xEcAkJFbxturzT3El2TkHfXcuZ6gvLWN3trL+PQeAOzOLaeowpz/CsoVlIuIiIiIdChv+Xpeec1Rl3JC/SZvddsreTPluZ25A7tXXB8zME+fDEYt7PoKPrgZ/jEY3rgMNr4DNS3rNN5ZedeTN1e67uUtYf9622G2ZpuZ9S3ZJc3OJcMw6oLyAJev94gMoX+i2Tx67YEiQN3XQUG5iIiIiEiH8m6JVuNyU15z9A7qJZV1Td68kqK95euduNFbfXF94LoFcMsamHknJA2D2hrYPh/evQ4eHgTfPgzHuIDR2R2ryZuXtwP7a8vryv0Lymt8JepHKqpwUlPrBurmTiCN7xMHwNp9Zgl7Xfd1BeUiIiIiItIBIkLsRHh69BxrWzTfHuVh9cvXPZnyzl6+fqSEgXDq7+Hm5XDj93Dy76BHf3BWwNd/hS/u7pKBeWmVk/35FWzLKQX8D8orjrigsyWrpMnjD3n2KO8R4SDUHvjeUBP6mCXsaw8U4ap1c9jThC46TEG5iIiIiIh0EF8J+zGy3d7y9dh6mfJEb6a8K5SvNydlBJz+Z/jlWjjzQfO+JY/B53/qUoH5vxZsZ/Q9Czjl4a8BGJIS5bvo0hzvtmheUwaYTeG2ZDcOymtcbv7y8RYABiRFNXo8ELyZ8nX7i5i/KYf88hoSIkMYkRoT2IEFkIJyEREREZEO5t2rvOAY26L5MuX1g3JP+XuXy5Q3xWKBKTfA2f80v172JHx6W5cJzD9Yn+W7bbNauGxS4y30jjQwKdJ3e0KfOGYOTQYaZ8oNw+AP76xnSUY+kSE27j1vZDuNum2GpkQTEWKjtNrF3z7ZCsCVU/p26y2Xu2+NgIiIiIhIgCR49yo/Vvm6Z5/y+o3ekgOQKa91G3y2KYfpgxMbZO07zMSfgtUOH/0aVvwfuF3mlmrWzptjLKqo8e3XvfrOWcRFhGCzWo75vLiIEBKjQsgrq2HuqFSGpUYD+Jq+eT30+XbmrcvCbrXw1JUnMKpXbPt/E61gt1kZ0zuWZbsLyCmpIsRu5copfQM9rIDqvLNYRERERKST8gXlfmfKG68pL6lyUe06eqO49jJvbSY3v76GP8/b1CGv16QTroXznwAssOoFeOtKqGp6HXVnsP5gMQD9EiJIiAr1KyD3umJyX0akxnDB+F4M95R978kv93Uyf+X7vTz9zS4A/n7RaE4dktTOo28b77pygAvGpQVFA7pAUlAuIiIiItLBvOXr+cdYU+7bp7xepjw23IHDZvHr+e1lx2GzCdlX2w5T43J3yGs2afyVcNGzYAuB7Z/A86dD3s7AjacNNni2BBubHtfi5/5m9hDm/+pkkqJDSYwKJTk6FMOAbTmlfL45h7s+3AzAb2cP4QcnprfjqNtH/aD8J9P7B3AkwUFBuYiIiIhIB0uJMYPy7zJyj5rtbqrRm8ViISHSuy1ax5SwHyo2O2SXVbtYubegQ16zWWN+CD/+FKLTIG8HvDAb9i8L7JhawZspH9M7rs3nGpFmZsvfWLGfX76xFsOAyyal84vTBrX53MfDtEEJTO4fz7XT+jGsZ/dt8OaloFxEREREpIOdOzaNhMgQdhwq4x+fb2/2uKb2KYe6/aZzO2hdebYnKAf4cuvhDnnNo+p9Ivz8W+h1IlQWwsvnw6r/QPHBTtEEzjAM1h8sAmBs77av9faWsL+z+iDVLjenDUvmL+ePwmLxvyS+I0WE2Hnr51O5J0iazwWagnIRERERkQ6WGBXKQ5eMAeC5xXv4bmdek8d5M+X1y9fN53u3VGs+KK9xubn9vQ3c9Npqat1tC1QPldQF5V9vD4KgHCAqGa75CIbMBVcVfPwbeGQk/HMovHEZLHoYdn1lBu1BJqekitzSamxWCyPT2h6U199ObGzvWJ64fDx2m0K9zkI/KRERERGRADh9eApXTDa3wLruvyv587xNHCioaHBMU43eoK7ZW3P7nLtq3fzqzbW8seIA8zfmNOrM3RKGYTTIlO/JK2d3blmrz9euQiLg0ldhxh3QcwxYbFB2CLbPh6/+Cq9cCA/2g5fOgfxdgR6tz/oDZun6kJRowkPavhXYpP7xhNqt9E+M5IVrJxIRok22OhP9tEREREREAuTOs0ewN7+cJRn5vLJsH6+v2M+5Y1K5YcZABiVFUV5jrjdvlCk/Svl6rdvgt2+v59NNOb77dhwqbfWWWMWVTqo9zd1O7NuDVfsK+WrbYQYkRbXqfO3OZocZt5kfzkrI3gCZq+s+CvfA3sXwzHSYfR+ceF3At1Jrz9J1gJSYML677TSiw+zder/vzkpBuYiIiIhIgISH2Hj1usks31PAU9/sYtGOXOaty2LeuixOGpTgOy46rOGf7UmeTHnuEeXrbrfB7e9t4APP/tRDe0azOauEHYdan9n2ZsnjI0OYOzrVF5T/9OQBrT7nceMIhz6TzQ+vgj3w4S1mYD7/d7DpXTjnUUgeFrBhbvAG5a3ovN6c7r6tWGem8nURERERkQCyWCxMGZDAyz+ZxMe3TOfsMalYLLAkIx+AqFB7o/XB3kx5Xr1MuWEY3P3hZv636iBWCzz2o/FcOtHcDmvnodJmX7+s2nXU8vYcz3rynjFhvv2uV+0rxOgEDdUAiO8PV38IZz4IjkjY/72ZNf/qr+CsOvbz/VDtqmXV3gJ25ZYdc8u47OJKNhzwdl5vn0y5dG7KlIuIiIiIBIlRvWJ58vIJ7Mkr59lFu3h3dSZj0xsHbt5M+fZDpeQUV5ESE8r9n2zllWX7sFjgnz8cy9ljUlm22wzsvfuM1+eqdfPmygM8snAH+eU1PHTxGH44sfGe1jmeTHnP2DDS4sIAs4lcRU0tkaGdJJywWmHKDTDsbDNbvuMzsxHcpvfg3Eeh/yltOv3jX+7kya/NNetWC6TFhdM3PgLKrOQs2cuA5Bj6JURQUuXiptdWU1rtIj0+nCEp0e3wzUln10n+LxIRERER6T76J0by94vG8OdzRmBvYv3zhL5xDEyKZFduOde+uIKTByfy/Hd7AHjgotFcOL43gC/oO1BQSUWNy9cAzDAMLn9+OSv21O05/tDn25g7uifRR6xfrx+UhztshNis1NS6Kap0dp6g3CsuHS57E7Z8AJ/eBgW74L/nwrgrYM5fISK+Vaddtdfs8G63WnC5DQ4WVnKwsBKwsuSzHY2OH5ISxQvXTMShDumCytdFRERERIJWRIidEHvjP9lD7TZe+vEkkqJD2ZZTynOLzYD8vvNHcunEPr7j4iNDfNun7ay3rrys2uULyO86ZwT9EyPJK6vhmW8bdyj3BeUxYVgsFmIjzKC9qKLpzu9Bz2KBkRfAL1aYTd+wwLrX4NHR8MpF8O1DsGcR1JT7fco9eeaxb98wlRV/Op23b5jKAxeOZHYvN2eNSmFUrxiiPBcwZg1P5p0bp5EeH3EcvrluYMPbULgv0KNoV53s0paIiIiIiACkx0fw4rUTufT/vqe8ppY7zhrG1VP7NTpucHI0eWX57DhU6mssVlBuBtThDhs/md6fXj3C+fkrq3l+8R6umNyXtLhw3/PrrykHiAt3kFtaTXGF8/h+g8dbWCyc8y8Y+yP46NdweDPs+tL8AHN7tdQx0GcqDDwNBs0yA/ojlFY5OexZ2z8wOYqYMAfJ0WGM6xVNeM56zjprLA6HA8MwqHTWaruy1nK74au/wHf/gqTh8NMvIDRIdgBoI80IEREREZFOalSvWOb/6mRyiquYPCChyWOGpETx/e58dh6uy5Tne4Ly+Egziz5nRAqT+sezYk8BD3++nUcuHec7tn75OkCcN1Ne2cmDcq/0SXDDd3BoE+xfBgeWwf7lUHIQstaaH8ueMgPzs/8J8Q27znuz5IlRoY22rqvPYrEoIG+NWicU7DYD8q0fmfcNOxscXafSQLNCRERERKQT65sQSd+EyGYfH+xZV76jXgf2gjIzKE/wlLZbLBbuPHs45z2xhPfXZvLjk/oxpnccUC9T7gnKY8PN5xR19kx5fVarmRVPHQOTrzfvKzoAB5bD3u9g3euw6yt4airMuhcm/9yXNfcG5QMSm/8ZiJ8K98KhzXB4a91H3g5we+aaLQTO+7dZ3dCFKCgXEREREenChvY0g/L6a8oLjsiUA4zpHceF43vx/tpM/vrJVt66fgpVTjfFnox440x5J11T7q+4dPNj9CUw7Rb4+Dew51v47DbY/Q2c/yREJrAr1xOUJykob7X9y8wt6vYubvrxkCjoORpm3QN9pnTo0DqCgnIRERERkS5sSLIZlGcWVVJW7SIq1N6ofN3r92cMZf7GbFbsKWDBlkO+7u0RITaiPY3K4sLNoLzTrylviYSBcPUHsOJZWHAn7PgUHh8HU24k59B0wOyYLy1QVWyWo697HfYtMe+z2iF5OCSPaPg5prdZzdBFKSgXEREREenCYiMcJEeHcri0mp2HShnfpwcF5WZjsoQjgvK0uHB+dvIAnvg6g7/P38p9548CzCy5xVOu7c2UF3eVNeX+sljMsvU+U2HejeYa9G8f5D4e5STHCfStvhxyKiEyCSKaXt/frbiqofigWZJetA+K9ptd04v2mZ8r8uqOtdhg/JVwyu/N6oRuRkG5iIiIiEgXNyQlmsOl1ezwBOX5Zd5MeWijY2+YMZA3Vx5gb34F/1po7rHt7bwOEBvRBdeUt0TqGPj5Ytj6Ica3DxJ2eAvn25bC0qWwtO4we3g8p7lDseU/DVHJZrAemQS9ToR+08ER1vxrdDaGYWa9t31SF3SXZgPG0Z+XOMRcHz7mUojt3SFDDUYKykVEREREurh+iRF8lwEHCyuBuu7rR2bKAaJC7dw6ewh3vL+RdQeKgLr15FBXvt7l15QfjdUKIy/gUK8zuP7B57jAvpRre+7FWp4LFfmAgaWygGiA/dmNn++INNdG9xxtfqSMgoRBYOtk4ZlhmA3wvvorZK1p/LgjAuL6Qo++5ue4Pg1vh8d1+JCDUSf7qYuIiIiISEulxpr7jmcWmUG5t9Gbt/v6kX54Ym9eWrqHHZ7mcPUz5b5Gb901U17P7rxyNhgDKY0dw09unmHe6a6FigKcJTms+OpjJo8eiL2qCMpzzXLu3V+bWeT6e6KD2VncEQFWm1nO7ftsBasD4vs3XGedOARCOmAdu2GAq8rshH5guVmOHpkI9nBzPfjhzeZxjkiYeB2kja8LxCMSmtzbXRpSUC4iIiIi0sX1ijOD8uwic3uzprqv12e3WbnjrOFc++JK4MhMufmcbremvAm7PduhNWjyZrVBVBKExpEXvRtjxFngqLd/uWFAzgY4uMpcl56zydwGzFkOtUepPijYBRlfNLwvJBpiUiFtAvQ7ydxD3RFhfoREmIFySATYw8wMfuFe8+KAqxrcLrOxmtVmXizI2wEl2VBdCtXF5ueqEvOz+yg/a0ckTLgaTv6t+X1LiykoFxERERHp4lI9QXV2sbd83dvorfGacq8ZQ5OZMyKFBVsOMT69h+/+2HBlyr1257Zij3KLBVLHmh9ebjcUHzAz0obbzLYbtZ7PbnBWQn6GZ+/uLeZHRT7UlEJeqRlQb3jzaC/KMdd3H0toLKRPguRhUFFgvn7faWZAHt7j2M+XZikoFxERERHp4tI8mfKs4irKq11UOd0AxDdTvu711BUTyCuraZApj/WUr1c6a6ly1hLmsB2nUR8f23NK+cO7G/jdnCGcPNjM7FbUuCiudPrK/P21J88s7+/f1j3KrVaz3Pto+p/c8OuqYig7bDZVO7DM3Ou77BDUVJhZ95oKqK32HOwJyKPTIDrFzJxb7WbQ73ZCVAokDYXYdAiLhdAYCI2GMM9n79cqRT8uFJSLiIiIiHRxKTFhWCxQ43KTcdgMJEPsViJDjh5Q223WBgE5QHSoHasF3AaUVDo7XVD+zLe7WH+giNeX7/cF5Te/tobvMvJ47adTmNQ/3u9zecvXByRGHZexHlVYrPmROBgGz2r6mFoXOCvMj7C4rtXxvQvpujuwi4iIiIgIYAbgSVFmqfqmrGLA7LxuaUXm02q11JWwd7J15TUuN19sPQTAvvwKAAzDYMWeApy1Bnd9sAlXrduvc1W7ajlQYJ5jYFsz5ceLzW5mu6N7KiAPYgrKRURERES6gVRPCfumzBKg+SZv/ojrpHuVL92VR2mVC4D9BRUYhkFeWQ3lNbUAbMsp5fUV+/0614GCCtwGRIbYSIpufm2+yLEoKBcRERER6QbSPGXomzLNTHlbgvK6Zm+da6/yzzbl+G6XVbsoKK9hb355g2P+uWCHrzv90ezyNnlLimpVxYGIl4JyEREREZFuwNvEbHtOKWCWr7eWb6/yTlS+Xus2WLDlUIP79hVUsNezLnzqgARGpMZQXOnk4c+3H/N8e5raDk2kFRSUi4iIiIh0A2lxZqa8xrNmOv4o26EdS5wnU14c5OXrlTW1XPjUEs7592Ke+CqDgvIa4iIcnNDX3MJrf36Fb235gKRI7j1/JABvrtzPxoPFRz337twy3/NE2kJBuYiIiIhIN+DdFs0r4RjboR2Nb015ZXCXr3+0IYu1+4vYlFnCI1/sAGDW8BTfvuL7Cyp85ev9EiKZ2C+eC8alYRhw94ebcLub39vbu0e5MuXSVgrKRURERES6gdQjtjZrnzXlwZ0pf23ZPgDG94nDYTPXfV8wrhd9EyIAswO7N1PezxNc337WcCJDbKzZX8T7azObPbe3fH1gUgC2Q5MuRfuUi4iIiIh0A0dmytvWfT3415RvPFjM+oPFOGwWnr/6RCqdtRwqqeKEvvEUeBrU7S8or5cpNwP1lJgwbjl9MA98uo2/f7qN2SNTiAlzNDh3cYWTfE8zOGXKpa2UKRcRERER6QYSo0KxW+u6hLdHo7dgXlP+2nIzSz53VCoJUaH07hHBCX3jAegbbwbgm7NKKK1yYbFAuuc+gJ+c1J8BiZHklVXz+Bc7G517d565njwlJpTIUOU5pW0UlIuIiIiIdAM2q4We9UrY25QpDw/uNeUlVU4+WJcFwJVT+jZ63Fu+XuHZnzwtNpwwh833eIjdyl3njgDgpaV7yThc2uD53vXkAxJVui5tp6BcRERERKSbSIutK2FPiGp99/XYiOBeU/7+mkwqnbUMSYliYr8ejR6PDXcQHVaX4fYG6fXNGJrMrOEpuNwG93y4BcOoa/rmzZT3V+d1aQcKykVEREREuolUz7ZoDpuFmLDWl10H85ZohmHwqqfB2xWT+2KxWBodY7FYGgTifROaDq7vOmcEIXYr32Xk8eTXGb77vU3eBmg9ubQDBeUiIiIiIt1EqidT3iMipMlg1V/eLdFKq124PPueB4uVewvZebiMcIeNCyf0ava4vvF1AXW/JjLlAH0SIvjzOWYZ+z8W7OCd1QeBuvJ1dV6X9qCuBCIiIiIi3USaJ1PelvXkQIMse0mVq83na0/eLPkF49MadU2vr0+9QLzfUTLeV03pS2ZhJc98u4s/vruBkkqnL1OuzuvSHpQpFxERERHpJsalxwEwIi2mTeex26y+NdlFFcHT7C2vrJpPN2UDZun60fSt1229XzPl615/OGMoF4xLw+U2uO/jLVS73DhsFnr3CD/q80T8oUy5iIiIiEg3MaZ3HN/dNpOUmLBjH3wMcREOSqtcQbVX+durDuKsNRibHseoXrFHPbZPvaC8/u2mWK0W/vnDcZzYL55HFu4gv7yGgUlR2G3KcUrbKSgXEREREelGevc4egDqr7jwEA5QGTTN3txug9dXeBu89Tnm8SPSYogKtTMwKZLwRUC5pAAAHRBJREFUENsxj7dZLVw5pS/njUvj7VUHObFv467uIq2hoFxERERERFos1tOBPVj2Kl+0M5cDBZXEhNk5d0zaMY+Piwhh8R9m+hWQ1xcT5uC66f1bO0yRRlRvISIiIiIiLRbovcora2r554LtbDhYBMCry/YDcMkJ6X4H2j0iQwhztCwoF2lvCspFRERERKTFvHuVByoo/3RTNv/+KoPLn1vOF1sO8dW2QwBc7kfpukgwUfm6iIiIiIi0WJwnU14coEZvmYWVAJRVu/jpy6sAmDoggUHJ2jtcOpeAZsoXLVrEueeeS1paGhaLhXnz5jV43DAM7rrrLlJTUwkPD2fWrFns3LkzMIMVERERERGfuHBzb/JAbYl2qLSq0X1XTjn6NmgiwSigQXl5eTljx47lySefbPLxhx56iMcff5xnnnmG5cuXExkZyRlnnEFVVeP/AUVEREREpOP41pQHKFN+qKQagJtnDqRfQgRDU6KZPSIlIGMRaYuAlq/PnTuXuXPnNvmYYRg8+uij3HnnnZx//vkAvPzyy6SkpDBv3jx+9KMfdeRQRURERESknkCvKT9cYibqxvaO4zezhmCzWrBYLAEZi0hbBO2a8j179pCTk8OsWbN898XGxjJ58mS+//77ZoPy6upqqqurfV+XlJQA4HQ6cTqDYw/F7sT7nuu9F39pzoi/NFekpTRnpKU0Z44uKsQsui2qqAnIe5TjCcoTIuwY7lpc7g4fwlFp/nRvLfm5B21QnpOTA0BKSsMSlJSUFN9jTfn73//Ovffe2+j+BQsWEBER0b6DFL8tXLgw0EOQTkZzRvyluSItpTkjLaU507TsCgA7ucXlzJ8/v0Nf221AbokNsLBp5RIObujQl28RzZ/uqaKiwu9jgzYob63bb7+dW2+91fd1SUkJ6enpzJkzh5iYmACOrHtyOp0sXLiQ2bNn43A4Aj0c6QQ0Z8RfmivSUpoz0lKaM0d3uLSaB9Z/S2WthTPPnIvV2nGl43ll1biXfYvFAj8870zstuDb6Vnzp3vzVmz7I2iD8p49ewJw6NAhUlNTffcfOnSIcePGNfu80NBQQkNDG93vcDj0P0MA6f2XltKcEX9prkhLac5IS2nONC0h2gyE3QZUuS3Ehnbce5TvyUImRoUSHtb4b/9govnTPbXkZx58l5Q8+vfvT8+ePfnyyy9995WUlLB8+XKmTp0awJGJiIiIiEiYw0a4wwZAcQc3ezvs2Q4tJSa4A3IRfwQ0U15WVkZGRobv6z179rBu3Tri4+Pp06cPv/71r/nrX//K4MGD6d+/P3/+859JS0vjggsuCNygRUREREQEgLgIB5XFtRRV1tCHjuvf5N0OLSU6rMNeU+R4CWhQvmrVKmbOnOn72rsW/JprruGll17iD3/4A+Xl5Vx//fUUFRUxffp0PvvsM8LC9D+fiIiIiEigxYY7yC6u6vBt0Q55Oq8nK1MuXUBAg/IZM2ZgGEazj1ssFu677z7uu+++DhyViIiIiIj4Iy7Cs1d5ZcuD8sOlVXy4LovzxqWR3MKMtzdT3tLniQSjoF1TLiIiIiIiwS0uPASA4lYE5c8t2s1fP9nKBU8sYUuW/52qAQ6XeNeUKyiXzk9BuYiIiIiItIo3U15cUdPi5x4uNbPdWcVV/OCZpSzakev3cw+p0Zt0IQrKRURERESkVWK95eutWFNeVuUyzxHuoLymlt++vZ4al9uv5/oavSlTLl2AgnIREREREWkVb/l6a9aUl3qC8nvOG0FydCi5pdV8tjnnmM9z1brJL/OsKVemXLoABeUiIiIiItIqcW3IlJdWm0F5fGQol0/uA8B/l+5t9vhDJVUcLKwgv7wGtwE2q4WESAXl0vkpKBcRERERkVaJDfesKa9s+ZrysmozkI8KtXP55D44bBZW7ytkU2Zxo2Nr3QbnP7GEMx5ZxNr9hQAkRYVis1raMHqR4KCgXEREREREWiUuvA2Zck/5ekyYneToMOaOSgXgpSay5Xvzy8kpqaK8ppb7528F1ORNug4F5SIiIiIi0iqxrdyn3DAMX6O3qDA7ANdM6wfAh+uzfGvGvbZll/puHyioBCBZTd6ki1BQLiIiIiIirRIX4dmnvMKJYRh+P6/a5cblNo+PCjWD8gl94hjdK5Yal5u3Vh1ocPz2nMb7mCtTLl2FgnIREREREWkVb/l6Ta2bSmet388rqTIz6xYLRIbYPbctvmz5q9/vw1Vbtz3athwzU372mFTffcnRypRL16CgXEREREREWiUixIbDZjZba8m6cl/peogda71mbeeMSSU+MoSs4iq+2HrId783KL9iUh9OG5YMwLCe0W0ev0gwUFAuIiIiIiKtYrFYiPXuVd6CoNzb5C3as57cK8xh40cT04G6hm/l1S72F1QAMLRnNE9ePoHXfzaZ2SNS2jp8kaCgoFxERERERFrNt1d5C7ZFK6tu2OStviun9MVmtbBsdwHbckrYccjMkidFh5IQFUp4iI1pAxOxWLQdmnQNCspFRERERKTVvOvKi1uRKfc2easvLS6cOZ4s+Mvf72O7p3Rd5erSVSkoFxERERGRVotrxbZopZ5Gb9FhjiYf9zZ8e39NJiv2FAAKyqXrUlAuIiIiIiKt1po15UcrXweY3D+eoSnRVDpreX9dJgBDe8a0caQiwUlBuYiIiIiItFpr1pR7y9djmgnK62+P5t3+XJly6aoUlIuIiIiISKvFR5qZ8oOFlX4/x5cpb2JNudcF49N8QbvVAoOSo9owSpHgpaBcRERERERabXL/eAAW78jFWev26zl1jd6aXlMOEBFi51LP9mj9EyMJc9jaOFKR4KSgXEREREREWm18nx7ER4ZQUuVi1d5Cv55T1+it+Uw5wM9OGcDJgxO5ccagNo9TJFgpKBcRERERkVazWS2cNiwZgC+2HvLrOcdq9OaVHB3GK9dN5pITerdtkCJBTEG5iIiIiIi0yazhdUG54e3MdhRlnvL16KOsKRfpLhSUi4iIiIhIm5w8OIkQm5V9+RXsyi075vHeNeXN7VMu0p0oKBcRERERkTaJDLUzbVACAAu3HD7m8f6Wr4t0BwrKRURERESkzU4fngL4t668xM9GbyLdgYJyERERERFpM++68jX7C8kvq272OMMwfJlyrSkXUVAuIiIiIiLtIDU2nFG9YjAM+Gpb8yXsFTW1eHvBqXxdREG5iIiIiIi0k1l+lLB7m7zZrBbCHbYOGZdIMFNQLiIiIiIi7cIblC/akUeVs7bJY8qqzfXkUaF2LBZLh41NJFgpKBcRERERkXYxMi2GnjFhVDpr+X53fpPHlPi2Q1PpuggoKBcRERERkXZisViYNcJs+PbFlqZL2Ms8QXmUmryJAArKRURERESkHXm3Rvty62EMb0e3enyd15UpFwEUlIuIiIiISDuaOiCBiBAbOSVVbM4qafR4qW+PckdHD00kKCkoFxERERGRdhPmsHHK4CQAFjZRwl6q8nWRBhSUi4iIiIhIuzp9uGddeRNbo6l8XaQhBeUiIiIiItKuThuWjMUCm7NKyC6ubPCYL1OuoFwEUFAuIiIiIiLtLCEqlBP69ADgi62HGzzm7b4erfJ1EUBBuYiIiIiIHAezRphd2L/YcohqVy3PL97N97vyKa1WozeR+nR5SkRERERE2t2s4ck88Ok2vt+Vz+XPLWf1vkJC7FbSYsMANXoT8VKmXERERERE2t3ApCj6JURQU+tm9b5CAGpcbvbmVwBaUy7ipaBcRERERETancVi4YxRPQHoEx/Bm9dPISEyxPe4uq+LmPR/goiIiIiIHBe3nDaYwcnRzBqeTFxECA9dMobr/rsKgBitKRcBFJSLiIiIiMhxEhVq55ITevu+Pn14CvedP5Kdh8oYkRoTwJGJBA8F5SIiIiIi0mGuntov0EMQCSpaUy4iIiIiIiISIArKRURERERERAJEQbmIiIiIiIhIgCgoFxEREREREQkQBeUiIiIiIiIiAaKgXERERERERCRAFJSLiIiIiIiIBIiCchEREREREZEAUVAuIiIiIiIiEiAKykVEREREREQCREG5iIiIiIiISIAEfVBeWlrKr3/9a/r27Ut4eDjTpk1j5cqVgR6WiIiIiIiISJsFfVD+05/+lIULF/LKK6+wceNG5syZw6xZs8jMzAz00ERERERERETaJKiD8srKSt59910eeughTjnlFAYNGsQ999zDoEGDePrppwM9PBEREREREZE2sQd6AEfjcrmora0lLCyswf3h4eF89913TT6nurqa6upq39clJSUAOJ1OnE7n8RusNMn7nuu9F39pzoi/NFekpTRnpKU0Z6QtNH+6t5b83C2GYRjHcSxtNm3aNEJCQnj99ddJSUnhjTfe4JprrmHQoEFs37690fH33HMP9957b6P7X3/9dSIiIjpiyCIiIiIiItKNVVRUcPnll1NcXExMTMxRjw36oHzXrl385Cc/YdGiRdhsNiZMmMCQIUNYvXo1W7dubXR8U5ny9PR08vLyjvlmSPtzOp0sXLiQ2bNn43A4Aj0c6QQ0Z8RfmivSUpoz0lKaM9IWmj/dW0lJCYmJiX4F5UFdvg4wcOBAvv32W8rLyykpKSE1NZVLL72UAQMGNHl8aGgooaGhje53OBz6nyGA9P5LS2nOiL80V6SlNGekpTRnpC00f7qnlvzMgz4o94qMjCQyMpLCwkI+//xzHnroIb+e5y0E8K4tl47ldDqpqKigpKREv4zEL5oz4i/NFWkpzRlpKc0ZaQvNn+7NG3/6U5ge9OXrn3/+OYZhMHToUDIyMvj9739PWFgYixcv9mtyHzx4kPT09A4YqYiIiIiIiEidAwcO0Lt376MeE/SZ8uLiYm6//XYOHjxIfHw8F198Mffff7/fV5vS0tI4cOAA0dHRWCyW4zxaOZJ3Tf+BAwe0pl/8ojkj/tJckZbSnJGW0pyRttD86d4Mw6C0tJS0tLRjHhv0mXLp3EpKSoiNjfWrwYEIaM6I/zRXpKU0Z6SlNGekLTR/xF/WQA9AREREREREpLtSUC4iIiIiIiISIArK5bgKDQ3l7rvvbnKbOpGmaM6IvzRXpKU0Z6SlNGekLTR/xF9aUy4iIiIiIiISIMqUi4iIiIiIiASIgnIRERERERGRAFFQLiIiIiIiIhIgCspFREREREREAkRBeTf097//nYkTJxIdHU1ycjIXXHAB27dvb3BMVVUVN998MwkJCURFRXHxxRdz6NAh3+Pr16/nsssuIz09nfDwcIYPH85jjz3W4BzfffcdJ510EgkJCYSHhzNs2DAeeeSRY47vvffeY86cOSQkJGCxWFi3bl2jY5599llmzJhBTEwMFouFoqKiVr0X4p+OmjP1LVmyBLvdzrhx4445PsMwuOuuu0hNTSU8PJxZs2axc+fOBsfcf//9TJs2jYiICOLi4lr0/Yv/usJcOe+88+jTpw9hYWGkpqZy1VVXkZWV1bI3QvzWFeZMv379sFgsDT4eeOCBlr0R0iKdfd588803jeaM92PlypUtf0PEb5197gCsWbOG2bNnExcXR0JCAtdffz1lZWUteyMkqCgo74a+/fZbbr75ZpYtW8bChQtxOp3MmTOH8vJy3zG/+c1v+Oijj3j77bf59ttvycrK4qKLLvI9vnr1apKTk3n11VfZvHkzf/rTn7j99tt54oknfMdERkbyi1/8gkWLFrF161buvPNO7rzzTp599tmjjq+8vJzp06fz4IMPNntMRUUFZ555JnfccUcb3gnxV0fNGa+ioiKuvvpqTj/9dL/G99BDD/H444/zzDPPsHz5ciIjIznjjDOoqqryHVNTU8MPfvADbrzxxja8E3IsXWGuzJw5k//9739s376dd999l127dnHJJZe04V2Ro+kKcwbgvvvuIzs72/dxyy23tPIdEX909nkzbdq0BvMlOzubn/70p/Tv358TTzyxje+OHE1nnztZWVnMmjWLQYMGsXz5cj777DM2b97Mtdde27Y3RgLLkG7v8OHDBmB8++23hmEYRlFRkeFwOIy3337bd8zWrVsNwPj++++bPc9NN91kzJw586ivdeGFFxpXXnmlX+Pas2ePARhr165t9pivv/7aAIzCwkK/zint43jPmUsvvdS48847jbvvvtsYO3bsUcfidruNnj17Gg8//LDvvqKiIiM0NNR44403Gh3/4osvGrGxscf4DqW9dOa54vXBBx8YFovFqKmpOer5pX10xjnTt29f45FHHvHzO5TjoTPOm/pqamqMpKQk47777jvquaX9dba583//939GcnKyUVtb6ztmw4YNBmDs3LnTr+9Zgo8y5UJxcTEA8fHxgHn1z+l0MmvWLN8xw4YNo0+fPnz//fdHPY/3HE1Zu3YtS5cu5dRTT22nkUugHM858+KLL7J7927uvvtuv8ayZ88ecnJyGrx2bGwskydPPuprS8fo7HOloKCA1157jWnTpuFwOPx6HWmbzjpnHnjgARISEhg/fjwPP/wwLpfLr9eQ9tFZ543Xhx9+SH5+Pj/+8Y/9eg1pP51t7lRXVxMSEoLVWhfGhYeHA+bSUemc7IEegASW2+3m17/+NSeddBKjRo0CICcnh5CQkEbrblNSUsjJyWnyPEuXLuWtt97ik08+afRY7969yc3NxeVycc899/DTn/603b8P6TjHc87s3LmTP/7xjyxevBi73b9fT97zp6Sk+P3a0jE681y57bbbeOKJJ6ioqGDKlCl8/PHHfr2GtE1nnTO//OUvmTBhAvHx8SxdupTbb7+d7Oxs/vWvf/n1OtI2nXXe1PfCCy9wxhln0Lt3b79eQ9pHZ5w7p512GrfeeisPP/wwv/rVrygvL+ePf/wjANnZ2X69jgQfZcq7uZtvvplNmzbx5ptvtvocmzZt4vzzz+fuu+9mzpw5jR5fvHgxq1at4plnnuHRRx/ljTfeAOC1114jKirK97F48eJWj0E6zvGaM7W1tVx++eXce++9DBkypMnnac50Lp15rvz+979n7dq1LFiwAJvNxtVXX41hGK3+PsQ/nXXO3HrrrcyYMYMxY8Zwww038M9//pN///vfVFdXt/r7EP911nnjdfDgQT7//HOuu+66Vo9fWqczzp2RI0fy3//+l3/+859ERETQs2dP+vfvT0pKSoPsuXQyga6fl8C5+eabjd69exu7d+9ucP+XX37Z5DrtPn36GP/6178a3Ld582YjOTnZuOOOO/x6zb/85S/GkCFDDMMwjJKSEmPnzp2+j4qKigbHak158Dmec6awsNAADJvN5vuwWCy++7788ssm58yuXbuanCennHKK8ctf/rLR96A15R2jK8wVrwMHDhiAsXTp0pa/EeK3rjRnNm3aZADGtm3bWv5GSIt0hXlz3333GUlJSepb0cG6wtzJyckxSktLjbKyMsNqtRr/+9//Wv+GSEApKO+G3G63cfPNNxtpaWnGjh07Gj3ubXDxzjvv+O7btm1bowYXmzZtMpKTk43f//73fr/2vffea/Tt29evYxWUB4+OmDO1tbXGxo0bG3zceOONxtChQ42NGzcaZWVlzY6tZ8+exj/+8Q/ffcXFxWr0FiBdaa547du3zwCMr7/+2p+3QFqoK86ZV1991bBarUZBQYFf74G0XFeZN2632+jfv7/x29/+tsXvgbROV5k79b3wwgtGRESE/h7uxBSUd0M33nijERsba3zzzTdGdna276N+pvqGG24w+vTpY3z11VfGqlWrjKlTpxpTp071Pb5x40YjKSnJuPLKKxuc4/Dhw75jnnjiCePDDz80duzYYezYscN4/vnnjejoaONPf/rTUceXn59vrF271vjkk08MwHjzzTeNtWvXGtnZ2b5jsrOzjbVr1xrPPfecARiLFi0y1q5da+Tn57fjOyVeHTVnjuRPp1LDMIwHHnjAiIuLMz744ANjw4YNxvnnn2/079/fqKys9B2zb98+Y+3atca9995rREVFGWvXrjXWrl1rlJaWtu5NkSZ19rmybNky49///rexdu1aY+/evcaXX35pTJs2zRg4cKBRVVXV+jdGmtXZ58zSpUuNRx55xFi3bp2xa9cu49VXXzWSkpKMq6++uvVvihxTZ583Xl988YUBGFu3bm35myCt0hXmzr///W9j9erVxvbt240nnnjCCA8PNx577LHWvSESFBSUd0NAkx8vvvii75jKykrjpptuMnr06GFEREQYF154YYOg+O67727yHPWz4I8//rgxcuRIIyIiwoiJiTHGjx9vPPXUUw22cGjKiy++2OS577777mO+fv3vQdpPR82ZI/n7D5jb7Tb+/Oc/GykpKUZoaKhx+umnG9u3b29wzDXXXNPk6yv72b46+1zZsGGDMXPmTCM+Pt4IDQ01+vXrZ9xwww3GwYMHW/N2iB86+5xZvXq1MXnyZCM2NtYICwszhg8fbvztb3/TRZzjrLPPG6/LLrvMmDZtWku+dWmjrjB3rrrqKiM+Pt4ICQkxxowZY7z88sstfRskyFgMQ51rRERERERERAJBLfpEREREREREAkRBuYiIiIiIiEiAKCgXERERERERCRAF5SIiIiIiIiIBoqBcREREREREJEAUlIuIiIiIiIgEiIJyERERERERkQBRUC4iIiIiIiISIArKRURERERERAJEQbmIiEgXd+2112KxWLBYLDgcDlJSUpg9ezb/+c9/cLvdfp/npZdeIi4u7vgNVEREpBtSUC4iItINnHnmmWRnZ7N3714+/fRTZs6cya9+9SvOOeccXC5XoIcnIiLSbSkoFxER6QZCQ0Pp2bMnvXr1YsKECdxxxx188MEHfPrpp7z00ksA/Otf/2L06NFERkaSnp7OTTfdRFlZGQDffPMNP/7xjykuLvZl3e+55x4Aqqur+d3vfkevXr2IjIz8/3bunqWRLo7D8M9ksQhioVhoIdEiNuKQRiFgIxYBI9iIghAEmwgWiuI30EEhIrGxEwtBRES7FCZaKBaSJvENbCRNfCuCBDHqmC0WFsLu0z27h2zuqzxnMpx/eTPkqKenR8fHx2YGBQCgwhDlAABUqb6+PlmWpb29PUmSy+VSLBbT5eWlNjc3lUwmNT8/L0kKBAJaXV1VfX29crmccrmc5ubmJElTU1M6OzvT9va20um0hoeHFQwGdXt7a2w2AAAqRU2pVCqZPgQAAPhzxsfHlc/ntb+//8ve6Oio0um0rq6uftnb3d1VJBLR8/OzpB//KZ+enlY+n//5TDabVXt7u7LZrFpaWn6u9/f3q7u7W4uLi//7PAAA/Eu+mT4AAAAwp1QqqaamRpJ0eHgo27Z1c3Ojl5cXfX5+6u3tTa+vr/J4PL/9fSaTkeM48vl8ZevFYlGNjY1//PwAAFQ6ohwAgCp2fX2ttrY23d3dKRQKaXJyUgsLC2poaNDJyYkmJib0/v7+n1FeKBTkdruVSqXkdrvL9urq6v7GCAAAVDSiHACAKpVMJpXJZDQzM6NUKqWvry9Fo1G5XD+unNnZ2Sl7vra2Vo7jlK35/X45jqPHx0f19vb+tbMDAPCvIMoBAKgCxWJR9/f3chxHDw8Pisfjsm1boVBI4XBYFxcX+vj40NramgYHB3V6eqr19fWyd3i9XhUKBSUSCVmWJY/HI5/Pp7GxMYXDYUWjUfn9fj09PSmRSKirq0sDAwOGJgYAoDJw+zoAAFUgHo+rublZXq9XwWBQR0dHisViOjg4kNvtlmVZWllZ0dLSkjo7O7W1tSXbtsveEQgEFIlENDIyoqamJi0vL0uSNjY2FA6HNTs7q46ODg0NDen8/Fytra0mRgUAoKJw+zoAAAAAAIbwpRwAAAAAAEOIcgAAAAAADCHKAQAAAAAwhCgHAAAAAMAQohwAAAAAAEOIcgAAAAAADCHKAQAAAAAwhCgHAAAAAMAQohwAAAAAAEOIcgAAAAAADCHKAQAAAAAw5DuHpkaHCHJ9YgAAAABJRU5ErkJggg==", - "text/plain": [ +"image/png": +"iVBORw0KGgoAAAANSUhEUgAAA+UAAAIjCAYAAABlBbqXAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjkuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8hTgPZAAAACXBIWXMAAA9hAAAPYQGoP6dpAAEAAElEQVR4nOzdd3hUZfbA8e+09J6QhEAg9NBBOqgUQUREURTFhmJbRV17+9lYdV3X3lZ3XYoF66rYEAQUQboovUNIIKT3Pu3+/rhzJwlpkzYzCefzPHlIMndm3plMwpx7znuOTlEUBSGEEEIIIYQQQrid3tMLEEIIIYQQQgghzlQSlAshhBBCCCGEEB4iQbkQQgghhBBCCOEhEpQLIYQQQgghhBAeIkG5EEIIIYQQQgjhIRKUCyGEEEIIIYQQHiJBuRBCCCGEEEII4SESlAshhBBCCCGEEB4iQbkQQgghhBBCCOEhEpQLIYQQXuTpp59Gp9ORnZ3tkfu/4YYbSEhI8Mh9//Of/yQxMRG73e6R+2/LLBYL8fHx/Otf//L0UoQQQjSSBOVCCCHc7l//+hc6nY5Ro0a5dPxDDz2ETqfjyiuvrPXy48ePo9PpeOmll5q0nu+++47x48cTHR1NQEAA3bt3Z/bs2axYscJ5zKlTp3j66afZsWNHk+7DHZYsWYJOp3N++Pn50bt3b+68804yMjI8vbx6FRYW8sILL/Dwww+j11e+Pan6ePR6PXFxcZx//vmsXbu2Vdbxr3/9iyVLljR43KeffopOp+Pf//53rZfffvvtmEwmdu7c2cIrrJ3JZOK+++7jueeeo7y83C33KYQQomVIUC6EEMLtli5dSkJCAlu3buXIkSP1HqsoCp988gkJCQl89913FBUVtehaXnrpJS6++GJ0Oh2PPvoor776KrNmzeLw4cN8+umnzuNOnTrFggULvDoo1/ztb3/jww8/5K233mLs2LG88847jBkzhtLS0gav+95773Hw4E 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 +oLL7zQqMcghBDCcyRTLoQQwm2WLl1KTEwMEydOBHB2VP/000+x2Wy1Hj98+HB69uwJQHBwMNOnT2/REvYFCxbw8ccfM3ToUFauXMn//d//MWzYMM466yz279/f4PV///13MjMzueOOO6rth54+fTqJiYn88MMPAGRlZbFu3TrmzZtHly5dqt2GTqercbt79uxh/PjxJCQksHr1asLDw11+TJMnT6ZDhw7Ex8dz1VVXERQUxNdff02nTp2qHXf77bc3eFtffvklOp2u1n3/2rq/+uor7HY7s2fPJjs72/kRGxtLr169+OWXX+q9j5ycHIxGI0FBQbVevnDhQjp06EB0dDSjRo1iw4YN3Hfffdxzzz0A+Pv7O48tKSkhOzubsWPHoigKf/75p/MYHx8f1q5dS15eXoOPuzFGjhzJX/7yF3Jzc3n++eed1QxffPEFffv2JTExsdrzom0nqPq8VH0MBQUFZGdnM378eI4dO0ZBQUG1++vWrRtTp06tdS3a68RTI/WEEEI0njR6E0II4RY2m41PP/2UiRMnOvcCA4waNYqXX36ZNWvWcP755zu/n5+fz/Lly7nzzjurdWgfN24cX375JYcOHaJ3794tsrY5c+YwZ84cCgsL2bJlC0uWLOHjjz9mxowZ7Nmzp9bmY5rk5GQA+vTpU+OyxMREfvvtNwCOHTsGwIABA1xa04wZM4iJiWHlypV1Bqt1efvtt+nduzdGo5GYmBj69OlTbcwYgNFopHPnzg3e1tGjR4mLiyMiIqLOYw4fPoyiKPTq1avWy5vbPO6SSy7hzjvvRKfTERwcTP/+/QkMDHRenpKSwpNPPsm3335bI+DWAlpfX19eeOEF7r//fmJiYhg9ejQXXXQR119/PbGxsc1aH+DsOaA12AP1edm/fz8dOnSo9TpVGwlu2LCBp556ik2bNtXokl9QUEBoaKjz627dutW5DkVRgNpP9AghhPBOEpQLIYRwi59//pm0tDQ+/fTTaq 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 +8fZsiQIbz88st89NFHQMtmmHv06MHOnTs577zz6r3d7777joqKCr799ttqWxsaKvuvjVaF0rdv38YvWAghhEfInnIhhBCtrqysjK+++oqLLrqIyy+/vMbHnXfeSVFREd9++y0AJ06cYN26dcyePbvW42+88UaOHDnCli1bmrWu0tJSNm3aVOtlP/74I1BZlq6VS5/eSXz48OFER0fz7rvvVhtz9eOPP7J//36mT58OQIcOHTj33HNZtGiRs9u3Ris5rkqn0/Gf//yHyy+/nLlz5zqfG3ebNWsWiqLUegJEW/dll12GwWBgwYIFNR6Loijk5OTUex9jxowBqDFizRVaZr7q/SqKwuuvv17tuNLSUsrLy6t9r0ePHgQHB1f7uQUGBjbYLd5Vs2fPJjU1lffee6/GZWVlZZSUlNT5GAoKCli8eHGj73P79u3odDrncyqEEML7SaZcCCFEq/v2228pKiri4osvrvXy0aNH06FDB5YuXcqVV17Jxx9/jKIodR5/4YUXYjQaWbp0KaNGjXJ+f82aNTUCL4CZM2fWupe7tLSUsWPHMnr0aC644ALi4+PJz89n2bJlrF+/npkzZzrLmnv06EFYWBjvvvsuwcHBBAYGMmrUKLp168YLL7zAjTfeyPjx45kzZw4ZGRm8/vrrJCQkcO+99zrv74033uDss8/mrLPO4tZbb6Vbt24cP36cH374gR07dtRYn16v56OPPmLmzJnMnj2b5cuX18jGt7aJEydy3XXX8cYbb3D48GHnqK/169czceJE7rzzTnr06MGzzz7Lo48+yvHjx5k5cybBwcEkJSXx9ddfc+utt/LAAw/UeR/du3dnwIABrF69mnnz5jVqfYmJifTo0YMHHniA1NRUQkJC+PLLL2vsLT906BDnnXces2fPpl+/fhiNRr7++msyMjK46qqrnMcNGzaMd955h2effZaePXsSHR3d5Of8uuuu4/PPP+cvf/kLv/zyC+PGjcNms3HgwA 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 +lBuSNTHtzE8nVQ944DJEQF1sgWXz+mK5P7RvPg1D5senQSL88e7HUBeWPodDr6xYVwx4SefH7bGGeGuaTC2sA1XXfIsZ/89KZzDTq1A76+Tf185G0w8HKXr+rrLF9vRqZcExQNl74D87dC3xmg2NQO7a8OgM/nwu+L1Ey6ojT/vrzACcfIP1f7OuSVmJ3bE/o3MlMO6haRy8/qDMBrqw81+vpCCCHaLsmUCyHOKO7stF1VY8vXnXvKm1i+XtzM8nWAOMdYtK6nla4DdO8QxH/ntt+GVEG+RsotZufz2BK0THmjmrzlJcMnc8BSCj3Og6l/b9R9+joqM5qdKa8qqhdc+REcXg3LH1Bno+9bpn4AhHaB7uOh+wToMUltHtcGpTiC8twSM3a7gl5f/9+NpjR5O92dk3ry5R8nWX84m23HcxmR0DafOyGEEI0jmXIhzhDvrD3K+Bd/ISWn1NNLaVUZheX8kZJX62X3f76Tya/82rIBiovMjey+HtXsRm/NG4kGMOuszlw4MJabXCzdbU8CHRUGLZkpr+y87mKmPOcoLL4Qik5BVB+4YjEYGneSxa8lM+Wn6zUZ7vwd5v0EEx6DruNAb4KCFPjzQ/jyJnhtEBxe1fL37QZaUG61KxSUWRo8frdzP3nTR5rFRwRwxfB4AF5dJdlyIYQ4U0imXIgzxPe7TpGcU8pHW5J57MK+nl5Oq7n1w+3sPJHP6vvOpWd0ZUbSarPzzY5UrHaFY1kl9ItrfHlpc5gdI6ka2+itqSPRmtt9HaBLZAD/uubM7LId6KM+by2VKS8oszjHxDlfl3Y75CdDWS6UF9T8+HMpFKdDVG+4/hvwa3ywp5Xhl7s6Eq2xDEboMkr9mPAwmEsgeRMc+0Xdc559CD6+Ei78JwybB/q2kwvQytdB/T0Mb6C3wO6TzQ/KQc2W/2/7CTYezWHzsZxqo/2EEEK0TxKUC3GGyC9Vg7Rvd5zi4QsSMTRQitkWKYrCoXS1RPiP5P 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 +UttFIhhBBNJUG5EO3YvlOVAXFju69D9cZpof4mfI2udVTvHxfKOb06ALBmf2aj77cx0p37yRtuXGVyBMVWu9L8MtBGMLfV8vWfn4EdS0Gnh8sXQp8LPL0itwpuQqbcYrNz72c7KLOo+9B3nSxwBvWeLl/Xfp+1aQWetie1gD2phfgY9Ewb0LHeY7W1p+aXcTijyPn95JyWmbt+Mk8N7oN9jY2uKnIH7aTqsz/sZ9fJAj7ZmtK8G4zsAZOfVj//+Vl1dJoQQgiPaWPvEIUQriq32Ljv8x0ATO0f06SAsGpQ3tB+8tNNTIwG4JcDrR2UO8ahNbCfHKqXj7uz2VubLF/f8yX89or6+SVvQ79LPLseD6gsX3e90dvrqw+z82SB8+s9qQXO7R+enFMO3pcp/3SbGlie3z+mwY7q2li0tPwyDmcUO79/vIWC8qpN3rxphrtGO1GwO1V9baUXllNhbV4DQobfpM4sL8uFX//Z3CUKIYRohjb0DlEI0Rj/+PEAhzOLiQry5blLBzbpNqoF5SGNC8onOYLyP1LynPvaW0NaYWMy5ZV/8tzZ7K3NNXpL3Q7L5qufj/srDLnas+vxEC0oN9vsLgVAW5NyeXvtEQBuHJcAwI4T+VQ4TsoEe7j7uvb7nF1c4fYtHKcrNVv55s9TAMwZ2aXB4yv3lJdzqEqm/HhOy5Svp+R4535yTfhp248UBdLym9mwz2CEC55XP9/6H8g62LzbE0II0WQefYe4bt06ZsyYQVxcHDqdjmXLltV57F/+8hd0Oh2vvfaa29YnRFu1PTmPJRuPA/DiFYOc44Qaq+qe8g6NvI1OYf4kxgZjV+DXVhyPlu7ijHKoHIkG7t1XbrGqpfJenynPTYJld8B/p4C1DHpOhvOe8vSqPEbbxwsNZ8sLyy3c+9kOFAUuH9aZ+8/vg04H2cWVJ6QCfV3b/tFaIgJ9nL8DmUWe7cD+w640iiqsdIkIqHMUWlUdHd3XM4rK2ZdWuS0nt8 +uPA/DiFYOc44Qaq+qe8g6NvI1OYf4kxgZjV+DXVhyPlu7ijHKoHIkG7t1XbrGqpfJenynPTYJld8B/p4C1DHpOhvOe8vSqPEbbxwsNZ8sLyy3c+9kOFAUuH9aZ+8/vg04H2cWVJ6QCfV3b/tFaIgJ9nL8DmUWe7cD+w640iiqsdIkIqHMUWlUdHd3XM4rK2ZdWuS0nt8TctPndp0nJVStuukR6Z1AeWqWkXntdnmyJ0XY9JkGfC9Uxh4svhB0fqxG/EEIIt/LoO8SSkhIGDx7M22+/Xe9xX3/9NZs3byYuLs5NKxOibdualAvA+f1imNgnusm3E1ctU95w0Hs6LVu+phVL2NNdnFEOoNfrMOq1sWju31Pu462Z8vwUtaHbW8PVPeSKTX2jPmsh6D0bSHqS0aDHz6T+zBraV/7ksj2k5pfRJSKApy/uT5CvkW5Rgc7L/U0GjB7++et0OmJC1ZNr2u+Np3y67QQAV46IR99Ag0aAqEBffAx6FAXnXHNNSgtky711RrlGOyl6Tq8oRnaLAOBEXstUCTDtn9ChL5Rmw7LbYcl0yNzfMrcthBDCJR59hzBt2jSeffZZLr300jqPSU1N5a677mLp0qWYTN7XfEUIb6Tts65v7q8rqpaENzZTDpVB+a8HM7G2Url4ula+7uJJA3d2YN95Ip8jmcVVGr152V7VgpPw/b3wxlnwxwdqtqznZLj5Z5jzCfiHeXqFHqftAy8os/D9rlP8mZKH7bQmgd/sSGXZjlMY9DpevXKI8zoDO4U6j/F053WN9nuS7sFZ5RmF5WxPzkOnU6sKXKHX65zZcu33qU9MMNAy+8q9dUa55voxXZkzsgt/v3QgncPVNZ5sqaA8LB7+sh4mL1A7sidvgHfPhlVPgrll9uwLIYSon3e8S6iD3W7nuuuu48EHH6R///4uXaeiooKKisomNoWFapmbxWLBYmmB+Z6iUbTnXJ5799Jm+UYHmZr13EcFVP6JiAgwNvq2Bn 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 +1OcQV3fvwnNrvCxYPjuHaUOpI0xF8NZgubu6fc8Tsf0ALzzjXBfiZuGNetRkAOakO08AA10D692dv25HxAzaID9I4N9qrGmA3RMuXJOSU8sWwPk1/51bmHvlVNeAT0RjiyGo5vaP37E0KIM4QE5UK0ATmnBeG/J9ddwn7Ki/eURwX5EOxrRFFodoMibfRbsJ+RQF/Xi360PeUVTciU70kt4O/L91NSUf/e2uwi9eflbzI4Rxe5nd0O38yHikLoPEJtziQElZUlhzIqpyDY7Qr3fr6T9MJyuncI5O+XDXQGqaH+anVFYXlzG72pvzf+rTVXuxZhjhL23NOavWl7sZ+8qB//uGwgL10x2G1raglaU7r8Ugsfbk7mSGYx/1p7pPXvOKI7nDVX/XzN36ARfYGEEELUTYJyIdqA3OLTgvI6StjNVjvZjq7fjSnpdhedTkc3R4On5u4rb2pDu8pMeePfTL7582H+s+4Yn/9+ot7jsorVtUUF+3gu+/b7Qji2Foz+jrJ12a0kVBN6RwOwen8GhzOKAHj7lyOsO5SFn0nPv645q1qptxaUN7v7uiNT7u/GE1WVmfLTgnLHfO+ze0Vx1cguXjOyzlWh/iaiHVU42qjF1fszKWrmiROXnPug+nflxGY4/FPr358QQpwBJCgXog3QMuVamfbWOoLyzKJyFEUd+xXppfsjEyLVN5DN3Vee7txP3rgy/eZ0X9e6T/+Rkl/vcc795EEeKl3POQqrnlQ/n7JAuq2LavrFhTC1fwyKAq+vOczGI9m8uvoQAM9cMoDE2JBqx7dUUK7tKXdnUK7tE6+65afUCkXl6u+yR6ciNNOiG0bwn+uGsere8fSMDsJstbNiT3rr33FIRxh1q/r5mmfUqhwhhBDNIkG5EG2A1uhtav8YQC2jLjPX3JOdVqXxmbfuj9T2lSdltUxQ3rGR89h9mrGnXMv0/VHP9gHwYJM3Szns+hw+mQOWUuh2Loy4xb1rEG3CPZPVOd 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 +eTFplGWpT50SuqF2lRafSO7s3QHkOxWxV6iDRF/2eIBECt22DBlkMAnDmy8f7Wvzp9MOk9Irh0YvP/OHYH3rXjRZVOnLVuyqrNP+7bnClf9gx8fgcY9bZAC42BgTNh8BwYNAuie/LtK6vILD5E4aFSQEGFyJEuOaE3T36dwcHCSl5bvo+fnjyg0TFVvu7r+v8nkBw2K9UuN66utKa8OVYrDD8Xhp0DOz6Dt66CLfNgYW844/5Aj67TCrOHMSFlAhNSJgBmL5bs8myWZC3hi31fsCJ7BZWuSnYV72JX8a5Gz48OiWZ6r+nM6D2Dk3qdRGxobEd/CyJBS0G5SACs3FtAQXkNcREOJg+Ib/R4enwEv5o1OAAjCy7ejHhheY0vSw5tXFO+eZ65hQ5A0jAzCB88B/pMAVvD8/ZNiATqukcrqBBpyGGz8svTBvOHdzfwzLe7uHxyn0Zbn6n7enCwe9aVu9xdOFN+JIsFhs6FC56G934K3z8BUSlw0i8DPbIuwWKxkBaVxg+G/IAfDPkBNbU1ZJdnk1mWSVZZFlllWb7bu4t3U1JTwqd7PuXTPZ9is9gYnTiaYfHDGBo/lKE9hjKoxyDC7d27QlC6LwXlIgHgLV2fNTwFh02ldM3xdncuqXKxPcfMVidHh2Jr7TZxB1fB+z83b0+6HuY+ZP7R1ow+8RENvg5TUCHSyIUTevHE1xnsL6jgle/38fNTBzZ43LtPudaUB5a3A3uX6r7urzE/gJKD8MU9sPDPYLWbHdqlXYXYQugb05e+MX0bPVbrrmVD3ga+OfANiw4uIqMog3W561iXu853jNViJT4snrjQOGJDY4kLjfN99AjrQe+o3vSP6096dDoOa/s3fBUJJAXlIh3M7TZ8QfmZI3se4+juLbZeRvzzzeZ7NnlAQutO5qyC/10DrioYfAac8fejBuQAfRMaBuXK9Ik05rBZ+eXpg/nd2+v5v0W7uXJK3wad2CtVaRIUvB3Yu0X5elNO+j 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 +jggg/4x6n/oF9MPwqrC3lo5UOc8/45vLfzPSpdlYEeasD4lSl/+eWXMQyDvXv3sm7dOvLz8zEMgyVLlnDyySfz8ssv43a72b17Ny+/bG4TcfXVVx/XgYt0RiXqvt56616HRQ+bt8991CwjFBERv9SVrytT3majL4GIeHjrKtjzLbx0FlzxLkSr67TIsVgtVs7odwan9zmdeRnzeHrd02SXZ3P30rv5x8p/cEb/MxiRMMJsFBfVm55RPXFYu34yy6/IoG/fvgCEhISQkpJC3759WbduHTExMcycORPDMKiursZisdCvXz8MQ7/wRY7kqnVTUVMLKChvsU3vwgc3AwZMucn8g0hERPzm3RLNqe7r7WPgaXDtx2amPGcjvDAbrngbkoYGemQinYLdaueSIZdwzoBzeHPbm7y5/U0yyzJ5Z0fDnQ2sFis9I3rSK7oXfaL7MDl1MtPSphEbGhugkR8ffkUGp556KgATJkzg448/5rbbbuOzzz7jrLPO4pRTzHUDGzduJD093fe1iDRUVu3y3daa8hbY+jG8+zMw3DDhGjhDy2RERFrKblWmvN2ljYfrFsArF0HhHnh6Gky63lx3ruVVIn4Js4dx7ahruXrk1azIWcHX+7/mQOkBDpYdJKssi+raarLKs8gqz2Jlzkre3fkuVouViT0n8uzsZ7FaukazxRal6x5++GHOP/98TjrpJIYNG8azzz7re+yll17izDPPbPcBinQVJZVmUB7msBJi7xq/QI67nV/A29eCUQtjfgTnPAoWS6BHJSLS6Tg8mXLtU97O4geYgfkHN8POBbDsKVj/Bpz6R5h4Hdh0EV7EH1aLlSmpU5iSWrfVoNtwk1eZ5+vkvqNwB99lfkdGUQZWrF0mIIcWBuVjx45l79695OfnN9qL/He/+x0xMTHtOjiRrsS7nlxN3o4trKYA6+r/wBd3gdsJIy6A85/U1jMiIq3k7b6ufcqPg6hks3Q940tYcCcc3gKf3QYrn4 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 +Y70Eo9mfKYcAdnj0kFmi9h9zZ565cQSXqPCODYJezeJm9xESFYj5JRFxER6Y4cNm+mXOXr3VZUEoy4wLy96oWADkWkNfI8fbOUKZeA8G6LllNcF5Q//e0uat0GJw9OZGx6nO9+b5b4yDXcgebdDi06zMH0wYlEh9k5XFrNyr0FjY71BuV94iNIjzeD8mM1eyso926HFjwl+yIiIsHC7uu+rkx5tzbxp+bnje+Y68tFOhHvmnIF5RIQR5av5xRX8c6qgwDcctrgBsfGhJlBafCVr3sy5WF2Qu02zhjZE4BPNjYuYT/gCcD7JETQN8GTKT9GUO4tX9d6chERkcbquq8rU96tpU+ClNHgqoJ1rwd6NCIt4i1f72qN3hSUdxIpvqC8GsMweHbRbmpq3UzqF8+k/g0bpQVv+bo5Hu/46krYcxp1it9XYDZ665sQSZ/4uvL14kon//luD59uzGZ/fkWDZoD1y9dFRESkobrydWXKuzWLBSb+xLy98gXQRRrpRLpqozd1w+okenq6r1c6a9mbX8HrK/YB8IvTBjU6NnjL183xeMd30sBEYsMd5JVVs2JPAVMHJgBQ5azlUIn5P1zf+Ahfmcre/HJueWMti3bk+s4ZHWpneGoMI9JiOFhoNoJT+bqIiEhj3vJ1p9aUy+gfwoK7oGAX7PkGBp4W6BFJJ2UYBre/t5HYCAe3zx1+XF+r1m34lqsmRnetJJwy5Z1EmMPm297s/k+2UuV0M6Z3LCcPTmx0rLd8Pfgy5d7ydXN8IXYrZ4w0u7B/sjHLd5y3TD06zE5chMNXvr45q4RFO3IJtVsZ1SuGEJuV0moXK/YW8NLSvXyx9RAAPSK71v+kIiIi7cFbvu5S93UJjYJxl5m3V6rhm7Reblk1b648wP99u/u4JwTzy6txG2axR3wXq4xVprwT6RkTRnGl0xd8/mLmICyWxl3GY8I9mfKgW1PuKV+vt13Z2WPS+N+qg3y2KYd7zh2J3Wb1NXnrmx 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 +PM1tXPlIeH2HxrX99bc/D4DLoNlu/O54Xv9gRvFj+2F/z4U0ifbG6P9soFsP3TQI+qS1i8Mw+ALdklAR5JQxVOM6YIP0qjN6+esUffnrmu0Vv3bAytoLybiAyx+UpLckurMQzDt5b88kl9SI01g+Ywh43RvWIB2HCwyO/z17jcXPXCcq57aSUF5TWAuc4kt8y8rUy5iIiItLcfTerDn84aTrjDxu68ch7+fDsnPfAVV/9nBR+tz6LKWdvoOVuzS6hxuYmLcNA/MbLBYxd6Stg/3pDd5HMD6U/zNvGXj7ewYk9BoIfSvIh4uGoeDDkTXFXw5hWw5pVAj6pTq3LWsnZ/EQC7Dpe1+aJMcYXzmBeu/OXLlPtRct7TkyBsblu0cjV6k+7AYrGQHFO3rnzRzjxW7ysk1G7lxhkDGxw7Jt0Mytc3s+ajKZ9vzmHxzjy+3HaYC59awq7cMg54StfDbQax4Y52+k5ERERE6vzslAGsvHMWD10yhkn94nEbsGhHLre8sZZJ93/BPR9u9q1XBViz31O6nh7XqLnclP4JpMWGUVrl4suthzv0+ziWPE9foObW5AaNkAi49FUYdwUYtfDhL2DxP7VdWiut2VdIjaeZYXlNLYea2d7YX1f9Zzlz/rWoXdan+7NPuZe3arf58nUF5dJNJEXV7VX+iGdNyhWT+5JyxJ6g43rHAbDe0wRlb145cx75lteXN99N89Vl+wCwWy3sy6/gvH9/xx/f2wBAYuMtR0VERETaTVSonR+emM7/bpjKN7+bwS2nDSItNoySKhcvLd3LJU8vJavITBas8WQdJ9RbT+5ltVp82fLjVcL+7KJdzPzHN2QW+Z+tdLsNSirNnXE2ZgZ5UA5gc8D5T8JJvza//vI+WP1SIEfUaX2/O7/B17tyy1p9Lmetmw0HiymtdrHIUxLfFuWe7uv+BOXeTPnhZi4qeM+lfcqly/Nmyv+36gDrDhQR5rByw4wBjY4bkx4HwPZDpVQ5a3 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 +POOrGFZw1e3optf3S5oDw0NJTQ0MadwB0Oh/4BbqPBKTFAFrWePQuumtrP7/dU77+0lOaM+EtzRVpKc0Zayt854wAe/dF4XO41fL75EDe8tpZfzxrC/y3eA8D9F4wmPCyUxKgwoJSiqlq/zlvqaYTVIyoMhyc9ml9e0znnceIAGDoXts/H8f2jcNGzgR7RcdfS3zkr9hYBcNKgpEbPG5gUxcbMYvYVVjE0Lc7vcxZWmnMoOTrMr7HEOhzMHhnO7JFpABwoqCA8xObbPq01HA7oEeGgsMJJQWUtybHmOKpqzdgiKqxr/W5uyfeifcrFb/Wvxo1Lj2NUr9gAjkZEREQk+DhsVh6/bDwzhiZR5XTzwKfbMAz4wQm9mTs6FYCEKDNDWODntmjeLdFiwh0kRZvPzSv1f5/zoHPqH8zPG9+GvIzAjiXIFFc62ZxlriefUq/Jm9dAz9/ju1q4V3n9Jm+tkR4f0aaA3CvFs1d5Tr29yr2N3qK66R7lEOCgvKysjHXr1vn2H9+zZw/r1q1j/35z/+uCggLWrVvHli1bANi+fTvr1q0jJycnUEPu1gbWC8q1DZqIiIhI00LtNp658gSmebaz6psQwd3n1ZUhJ0SagbV3d5tj8W6JFhNm9wVGJVUuql2dbK9yr7TxMGQuGG5Y9HCgRxNUVuwpwG1A/8RIX7fy+rwd2Fu6LZp3j3Jv6XmgNNWBvcy7JVo37r4e0KB81apVjB8/nvHjxwNw6623Mn78eO666y4APvzwQ8aPH8/ZZ58NwI9+9CPGjx/PM888E7Axd2d9EyLplxBBv4QIzhmTGujhiIiIiAStMIeNF66ZyIMXj+bN66c0yAImeAJrf/Yqr3Ubvm2oYsIdxIY7cHi6dOW1YK/zoDPjNvPzxv9B3s7AjiWILPNshdZUlhxgYHLrtkXz7lHuzVQHSkq0NyivuyBVtyVa982UB/Q7nzFjBoZhNPv4tddey7XXXttxA5KjctisfP6bU3C7zX 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 +e5MuXSVgrKRURERES6gdQjtjZrnzXlwZ0pf23ZPgDG94nDYTPXfV8wrhd9EyIAswO7N1PezxNc337WcCJDbKzZX8T7azObPbe3fH1gUgC2Q5MuRfuUi4iIiIh0A0dmytvWfT3415RvPFjM+oPFOGwWnr/6RCqdtRwqqeKEvvEUeBrU7S8or5cpNwP1lJgwbjl9MA98uo2/f7qN2SNTiAlzNDh3cYWTfE8zOGXKpa2UKRcRERER6QYSo0KxW+u6hLdHo7dgXlP+2nIzSz53VCoJUaH07hHBCX3jAegbbwbgm7NKKK1yYbFAuuc+gJ+c1J8BiZHklVXz+Bc7G517d565njwlJpTIUOU5pW0UlIuIiIiIdAM2q4We9UrY25QpDw/uNeUlVU4+WJcFwJVT+jZ63Fu+XuHZnzwtNpwwh833eIjdyl3njgDgpaV7yThc2uD53vXkAxJVui5tp6BcRERERKSbSIutK2FPiGp99/XYiOBeU/7+mkwqnbUMSYliYr8ejR6PDXcQHVaX4fYG6fXNGJrMrOEpuNwG93y4BcOoa/rmzZT3V+d1aQcKykVEREREuolUz7ZoDpuFmLDWl10H85ZohmHwqqfB2xWT+2KxWBodY7FYGgTifROaDq7vOmcEIXYr32Xk8eTXGb77vU3eBmg9ubQDBeUiIiIiIt1EqidT3iMipMlg1V/eLdFKq124PPueB4uVewvZebiMcIeNCyf0ava4vvF1AXW/JjLlAH0SIvjzOWYZ+z8W7OCd1QeBuvJ1dV6X9qCuBCIiIiIi3USaJ1PelvXkQIMse0mVq83na0/eLPkF49MadU2vr0+9QLzfUTLeV03pS2ZhJc98u4s/vruBkkqnL1OuzuvSHpQpFxERERHpJsalxwEwIi2mTeex26y+NdlFFcHT7C2vrJpPN2UDZun60fSt1229XzPl615/OGMoF4xLw+U2uO/jLVS73DhsFnr3CD/q80T8oUy5iIiIiEg3MaZ3HN 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", +"text/plain": [ "
" ] }, @@ -401,7 +592,7 @@ } ], "source": [ - "result = await team.run(\"Write a financial report on American airlines\")\n", + "result = await team.run(\"Write a financial report on American airlines\", termination_condition=StopMessageTermination())\n", "print(result)" ] } @@ -422,9 +613,9 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.12.6" } }, "nbformat": 4, "nbformat_minor": 2 -} +} \ No newline at end of file diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/literature-review.ipynb b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/literature-review.ipynb index cefb6e6f49ce..73b4943d3429 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/literature-review.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/literature-review.ipynb @@ -18,7 +18,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 1, "metadata": {}, "outputs": [], "source": [ @@ -72,6 +72,30 @@ "\n", " url = \"https://www.googleapis.com/customsearch/v1\"\n", " params = {\"key\": api_key, \"cx\": search_engine_id, \"q\": query, \"num\": num_results}\n", + "\n", + " response = requests.get(url, params=params) # type: ignore[arg-type]\n", + "\n", + " if response.status_code != 200:\n", + " print(response.json())\n", + " raise Exception(f\"Error in API request: {response.status_code}\")\n", + "\n", + " results = response.json().get(\"items\", [])\n", + "\n", + " def get_page_content(url: str) -> str:\n", + " try:\n", + " response = requests.get(url, timeout=10)\n", + " soup = BeautifulSoup(response.content, \"html.parser\")\n", + " return soup.get_text()[:max_chars]\n", + " except Exception as e:\n", + " print(f\"Failed to get page content from {url}: {e}\")\n", + " return \"\"\n", + "\n", + " return [get_page_content(item['link']) for item in results]\n" + ] + } + ] +} +" params = {\"key\": api_key, \"cx\": search_engine_id, \"q\": query, \"num\": num_results}\n", "\n", " response = requests.get(url, params=params) # type: ignore[arg-type]\n", "\n", @@ -150,6 +174,30 @@ ")" ] }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Defining Agents \n", + "\n", +"\n", + " return results" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [], + "source": [ + "google_search_tool = FunctionTool(\n", + " google_search, description=\"Search Google for information, returns results with a snippet and body content\"\n", + ")\n", + "arxiv_search_tool = FunctionTool(\n", + " arxiv_search, description=\"Search Arxiv for papers related to a given topic, including abstracts\"\n", + ")" + ] + }, { "cell_type": "markdown", "metadata": {}, @@ -161,7 +209,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 4, "metadata": {}, "outputs": [ { @@ -182,8 +230,9 @@ "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-08T13:16:39.878814], tool_agent_for_Google_Search_Agent:\u001b[0m\n", "\n", - "[FunctionExecutionResult(content='[{\\'title\\': \\'AutoGen Studio: A No-Code Developer Tool for Building and ...\\', \\'link\\': \\'https://arxiv.org/abs/2408.15247\\', \\'snippet\\': \\'Aug 9, 2024 ... Abstract:Multi-agent systems, where multiple agents (generative AI models + tools) collaborate, are emerging as an effective pattern for\\\\xa0...\\', \\'body\\': \\'[2408.15247] AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems Skip to main content We gratefully acknowledge support from the Simons Foundation, member institutions , and all contributors. Donate > cs > arXiv:2408.15247 Help | Advanced Search All fields Title Author Abstract Comments Journal reference ACM classification MSC classification Report number arXiv identifier DOI ORCID arXiv author ID Help pages Full text Search open search GO open navigation menu quick links Login Help Pages About Computer Science > Software Engineering arXiv:2408.15247 (cs) [Submitted on 9 Aug 2024] Title: AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems Authors: Victor Dibia , Jingya Chen , Gagan Bansal , Suff Syed , Adam Fourney , Erkang Zhu , Chi Wang , Saleema Amershi View a PDF of the paper titled AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems, by Victor Dibia and 7 other authors View\\'}, {\\'title\\': \\'AutoGen Studio: A No-Code Developer Tool for Building and ...\\', \\'link\\': \\'https://www.microsoft.com/en-us/research/publication/autogen-studio-a-no-code-developer-tool-for-building-and-debugging-multi-agent-systems/\\', \\'snippet\\': \\'Aug 2, 2024 ... Multi-agent systems, where multiple agents (generative AI models + tools) collaborate, are emerging as an effective pattern for solving\\\\xa0...\\', \\'body\\': \\'Your request has been blocked. This could be due to several reasons. Skip to main content Microsoft Microsoft 365 Teams Copilot Windows Surface Xbox Deals Small Business Support More All Microsoft Office Windows Surface Xbox Deals Support Software Windows Apps OneDrive Outlook Skype OneNote Microsoft Teams Microsoft Edge PCs & Devices Computers Shop Xbox Accessories VR & mixed reality Phones Entertainment Xbox Game Pass Ultimate Xbox Live Gold Xbox games PC games Windows digital games Movies & TV Business Microsoft Azure Microsoft Dynamics 365 Microsoft 365 Microsoft Industry Data platform Microsoft Advertising Licensing Shop Business Developer & IT .NET Visual Studio Windows Server Windows Dev Center Docs Other Microsoft Rewards Free downloads & security Education Store locations Gift cards View Sitemap Search Search Microsoft.com Cancel Your current User-Agent string appears to be from an automated process, if this is incorrect, please click this link: United States English\\'}, {\\'title\\': \\'Insights and Learnings from Building a Complex Multi-Agent System ...\\', \\'link\\': \\'https://www.reddit.com/r/LangChain/comments/1byz3lr/insights_and_learnings_from_building_a_complex/\\', \\'snippet\\': \"Apr 8, 2024 ... I\\'m a business owner and a tech guy with a background in math, coding, and ML. Since early 2023, I\\'ve fallen in love with the LLM world. So, I\\\\xa0...\", \\'body\\': \"You\\'ve been blocked by network security. To continue, log in to your Reddit account or use your developer token If you think you\\'ve been blocked by mistake, file a ticket below and we\\'ll look into it. Log in File a ticket\"}, {\\'title\\': \\'Multi Agents System (MAS) Builder - Build your AI Workforce\\', \\'link\\': \\'https://relevanceai.com/multi-agents\\', \\'snippet\\': \\'Mar 10, 2024 ... Easily build a multi-agent system. AI workers working collaboratively. No coding required.\\', \\'body\\': \\'Multi Agents System (MAS) Builder - Build your AI Workforce Recruit Bosh, the AI BDR Agent, and book meetings on autopilot. Recruit Bosh, the AI BDR Agent, and book meetings on autopilot. Register Learn more Product AI Agents Agent Teams AI Tools Integrations API Function Sales Marketing Customer Support Research Operations Agents Bosh the Sales Agent Inbound - AI SDR Outbound - AI BDR Lima the Lifecycle Agent Resources Blog Customers Documentation\\\\u200b Javascript SDK Python SDK\\\\u200b Templates Building the AI Workforce What is the AI Workforce? Enterprise Pricing Login Sign Up Product AI Agents Agent Teams AI Tools Custom Actions for GPTs API By Function Sales Marketing Customer Support Research Operations Agents Bosh the Sales Agent Inbound - AI SDR Outbound - AI SDR Resources Blog Documentation Workflows Javascript SDK Python SDK Templates Building the AI Workforce Enterprise Pricing Log in Sign up AI Agent Teams Build a Multi Agent System Create your own AI team that work collaboratively\\'}, {\\'title\\': \\'Crew AI\\', \\'link\\': \\'https://www.crewai.com/\\', \\'snippet\\': \"Start by using CrewAI\\'s framework or UI Studio to build your multi-agent automations—whether coding from scratch or leveraging our no-code tools and templates.\", \\'body\\': \\'Crew AI Get the Inside Scoop First! Join Our Exclusive Waitlist Home Enterprise Open Source Login Start Enterprise Trial crewAI © Copyright 2024 Log in Start Enterprise Trial The Leading Multi-Agent Platform The Leading Multi-Agent Platform Streamline workflows across industries with powerful AI agents. Build and deploy automated workflows using any LLM and cloud platform. Start Free Trial I Want A Demo 100,000,000+ 75,000,000 50,000,000 25,000,000 10,000,000 7,500,000 5,000,000 2,500,000 1,000,000 750,000 500,000 250,000 100,000 75,000 50,000 25,000 10,000 5,000 2,500 1,000 500 250 100 50 10 0 Multi-Agent Crews run using CrewAI Trusted By Industry Leaders The Complete Platform for Multi-Agent Automation 1. Build Quickly Start by using CrewAI’s framework or UI Studio to build your multi-agent automations—whether coding from scratch or leveraging our no-code tools and templates. 2. Deploy Confidently Move the crews you built to production with powerful tools for different deployment\\'}]', call_id='call_uJyuIbKg0XGXTqozjBMUCQqX')]\n", - "From: tool_agent_for_Google_Search_Agent\n", +"[FunctionExecutionResult(content='[{\\'title\\': \\'AutoGen Studio: A No-Code Developer Tool for Building and ...\\', \\'link\\': \\'https://arxiv.org/abs/2408.15247\\', \\'snippet\\': \\'Aug 9, 2024 ... Abstract:Multi-agent systems, where multiple agents (generative AI models + tools) collaborate, are emerging as an effective pattern for\\\\xa0...\\', \\'body\\': \\'[2408.15247] AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems Skip to main content We gratefully acknowledge support from the Simons Foundation, member institutions , and all contributors. Donate > cs > arXiv:2408.15247 Help | Advanced Search All fields Title Author Abstract Comments Journal reference ACM classification MSC classification Report number arXiv identifier DOI ORCID arXiv author ID Help pages Full text Search open search GO open navigation menu quick links Login Help Pages About Computer Science > Software Engineering arXiv:2408.15247 (cs) [Submitted on 9 Aug 2024] Title: AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems Authors: Victor Dibia , Jingya Chen , Gagan Bansal , Suff Syed , Adam Fourney , Erkang Zhu , Chi Wang , Saleema Amershi View a PDF of the paper titled AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems, by Victor Dibia and 7 other authors View\\'}, {\\'title\\': \\'AutoGen Studio: A No-Code Developer Tool for Building and ...\\', \\'link\\': \\'https://www.microsoft.com/en-us/research/publication/autogen-studio-a-no-code-developer-tool-for-building-and-debugging-multi-agent-systems/\\', \\'snippet\\': \\'Aug 2, 2024 ... Multi-agent systems, where multiple agents (generative AI models + tools) collaborate, are emerging as an effective pattern for solving\\\\xa0...\\', \\'body\\': \\'Your request has been blocked. This could be due to several reasons. Skip to main content Microsoft Microsoft 365 Teams Copilot Windows Surface Xbox Deals Small Business Support More All Microsoft Office Windows Surface Xbox Deals Support Software Windows Apps OneDrive Outlook Skype OneNote Microsoft Teams Microsoft Edge PCs & Devices Computers Shop Xbox Accessories VR & mixed reality Phones Entertainment Xbox Game Pass Ultimate Xbox Live Gold Xbox games PC games Windows digital games Movies & TV Business Microsoft Azure Microsoft Dynamics 365 Microsoft 365 Microsoft Industry Data platform Microsoft Advertising Licensing Shop Business Developer & IT .NET Visual Studio Windows Server Windows Dev Center Docs Other Microsoft Rewards Free downloads & security Education Store locations Gift cards View Sitemap Search Search Microsoft.com Cancel Your current User-Agent string appears to be from an automated process, if this is incorrect, please click this link: United States English\\'}, {\\'title\\': \\'Insights and Learnings from Building a Complex Multi-Agent System ...\\', \\'link\\': \\'https://www.reddit.com/r/LangChain/comments/1byz3lr/insights_and_learnings_from_building_a_complex/\\', \\'snippet\\': \"Apr 8, 2024 ... I\\'m a business owner and a tech guy with a background in math, coding, and ML. Since early 2023, I\\'ve fallen in love with the LLM world. So, I\\\\xa0...\", \\'body\\': \"You\\'ve been blocked by network security. To continue, log in to your Reddit account or use your developer token If you think you\\'ve been blocked by mistake, file a ticket below and we\\'ll look into it. Log in File a ticket\"}, {\\'title\\': \\'Multi Agents System (MAS) Builder - Build your AI Workforce\\', \\'link\\': \\'https://relevanceai.com/multi-agents\\', \\'snippet\\': \\'Mar 10, 2024 ... Easily build a multi-agent system. AI workers working collaboratively. No coding required.\\', \\'body\\': \\'Multi Agents System (MAS) Builder - Build your AI Workforce Recruit Bosh, the AI BDR Agent, and book meetings on autopilot. Recruit Bosh, the AI BDR Agent, and book meetings on autopilot. Register Learn more Product AI Agents Agent Teams AI Tools Integrations API Function Sales Marketing Customer Support Research Operations Agents Bosh the Sales Agent Inbound - AI SDR Outbound - AI BDR Lima the Lifecycle Agent Resources Blog Customers Documentation\\\\u200b Javascript SDK Python SDK\\\\u200b Templates Building the AI +in File a ticket\"}, {\\'title\\': \\'Multi Agents System (MAS) Builder - Build your AI Workforce\\', \\'link\\': \\'https://relevanceai.com/multi-agents\\', \\'snippet\\': \\'Mar 10, 2024 ... Easily build a multi-agent system. AI workers working collaboratively. No coding required.\\', \\'body\\': \\'Multi Agents System (MAS) Builder - Build your AI Workforce Recruit Bosh, the AI BDR Agent, and book meetings on autopilot. Recruit Bosh, the AI BDR Agent, and book meetings on autopilot. Register Learn more Product AI Agents Agent Teams AI Tools Integrations API Function Sales Marketing Customer Support Research Operations Agents Bosh the Sales Agent Inbound - AI SDR Outbound - AI BDR Lima the Lifecycle Agent Resources Blog Customers Documentation\\\\u200b Javascript SDK Python SDK\\\\u200b Templates Building the AI Workforce What is the AI Workforce? Enterprise Pricing Login Sign Up Product AI Agents Agent Teams AI Tools Custom Actions for GPTs API By Function Sales Marketing Customer Support Research Operations Agents Bosh the Sales Agent Inbound - AI SDR Outbound - AI SDR Resources Blog Documentation Workflows Javascript SDK Python SDK Templates Building the AI Workforce Enterprise Pricing Log in Sign up AI Agent Teams Build a Multi Agent System Create your own AI team that work collaboratively\\'}, {\\'title\\': \\'Crew AI\\', \\'link\\': \\'https://www.crewai.com/\\', \\'snippet\\': \"Start by using CrewAI\\'s framework or UI Studio to build your multi-agent automations—whether coding from scratch or leveraging our no-code tools and templates.\", \\'body\\': \\'Crew AI Get the Inside Scoop First! Join Our Exclusive Waitlist Home Enterprise Open Source Login Start Enterprise Trial crewAI © Copyright 2024 Log in Start Enterprise Trial The Leading Multi-Agent Platform The Leading Multi-Agent Platform Streamline workflows across industries with powerful AI agents. Build and deploy automated workflows using any LLM and cloud platform. Start Free Trial I Want A Demo 100,000,000+ 75,000,000 50,000,000 25,000,000 10,000,000 7,500,000 5,000,000 2,500,000 1,000,000 750,000 500,000 250,000 100,000 75,000 50,000 25,000 10,000 5,000 2,500 1,000 500 250 100 50 10 0 Multi-Agent Crews run using CrewAI Trusted By Industry Leaders The Complete Platform for Multi-Agent Automation 1. Build Quickly Start by using CrewAI’s framework or UI Studio to build your multi-agent automations—whether coding from scratch or leveraging our no-code tools and templates. 2. Deploy Confidently Move the crews you built to production with powerful tools for different deployment\\'}]', call_id='call_uJyuIbKg0XGXTqozjBMUCQqX')]\n", +"From: tool_agent_for_Google_Search_Agent\n", "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-08T13:16:49.739108], Google_Search_Agent:\u001b[0m\n", "\n", @@ -217,13 +266,23 @@ "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-08T13:16:50.542039], Arxiv_Search_Agent:\u001b[0m\n", "\n", +"3. Crew AI. (2024). The Leading Multi-Agent Platform. Retrieved from [Crew AI](https://www.crewai.com/).\n", + "4. Insights from Community Discussions. Reddit. (April 2024). \n", + "\n", + "This review highlights the emerging trends and significant tools in the no-code multi-agent AI space, indicating a shift toward more accessible AI system development.\n", + "From: Google_Search_Agent\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-08T13:16:50.542039], Arxiv_Search_Agent:\u001b[0m\n", + "\n", "[FunctionCall(id='call_HnNhdJzH3xCbiofbbcoqzFDP', arguments='{\"query\":\"no code tools multi agent AI systems\",\"max_results\":5}', name='arxiv_search')]\n", "From: Arxiv_Search_Agent\n", "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-08T13:16:52.486634], tool_agent_for_Arxiv_Search_Agent:\u001b[0m\n", "\n", - "[FunctionExecutionResult(content='[{\\'title\\': \\'AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems\\', \\'authors\\': [\\'Victor Dibia\\', \\'Jingya Chen\\', \\'Gagan Bansal\\', \\'Suff Syed\\', \\'Adam Fourney\\', \\'Erkang Zhu\\', \\'Chi Wang\\', \\'Saleema Amershi\\'], \\'published\\': \\'2024-08-09\\', \\'abstract\\': \\'Multi-agent systems, where multiple agents (generative AI models + tools)\\\\ncollaborate, are emerging as an effective pattern for solving long-running,\\\\ncomplex tasks in numerous domains. However, specifying their parameters (such\\\\nas models, tools, and orchestration mechanisms etc,.) and debugging them\\\\nremains challenging for most developers. To address this challenge, we present\\\\nAUTOGEN STUDIO, a no-code developer tool for rapidly prototyping, debugging,\\\\nand evaluating multi-agent workflows built upon the AUTOGEN framework. AUTOGEN\\\\nSTUDIO offers a web interface and a Python API for representing LLM-enabled\\\\nagents using a declarative (JSON-based) specification. It provides an intuitive\\\\ndrag-and-drop UI for agent workflow specification, interactive evaluation and\\\\ndebugging of workflows, and a gallery of reusable agent components. We\\\\nhighlight four design principles for no-code multi-agent developer tools and\\\\ncontribute an open-source implementation at\\\\nhttps://github.com/microsoft/autogen/tree/main/samples/apps/autogen-studio\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/2408.15247v1\\'}, {\\'title\\': \\'Collaboration of AI Agents via Cooperative Multi-Agent Deep Reinforcement Learning\\', \\'authors\\': [\\'Niranjan Balachandar\\', \\'Justin Dieter\\', \\'Govardana Sachithanandam Ramachandran\\'], \\'published\\': \\'2019-06-30\\', \\'abstract\\': \\'There are many AI tasks involving multiple interacting agents where agents\\\\nshould learn to cooperate and collaborate to effectively perform the task. Here\\\\nwe develop and evaluate various multi-agent protocols to train agents to\\\\ncollaborate with teammates in grid soccer. We train and evaluate our\\\\nmulti-agent methods against a team operating with a smart hand-coded policy. As\\\\na baseline, we train agents concurrently and independently, with no\\\\ncommunication. Our collaborative protocols were parameter sharing, coordinated\\\\nlearning with communication, and counterfactual policy gradients. Against the\\\\nhand-coded team, the team trained with parameter sharing and the team trained\\\\nwith coordinated learning performed the best, scoring on 89.5% and 94.5% of\\\\nepisodes respectively when playing against the hand-coded team. Against the\\\\nparameter sharing team, with adversarial training the coordinated learning team\\\\nscored on 75% of the episodes, indicating it is the most adaptable of our\\\\nmethods. The insights gained from our work can be applied to other domains\\\\nwhere multi-agent collaboration could be beneficial.\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/1907.00327v1\\'}, {\\'title\\': \\'Levels of AI Agents: from Rules to Large Language Models\\', \\'authors\\': [\\'Yu Huang\\'], \\'published\\': \\'2024-03-06\\', \\'abstract\\': \\'AI agents are defined as artificial entities to perceive the environment,\\\\nmake decisions and take actions. Inspired by the 6 levels of autonomous driving\\\\nby Society of Automotive Engineers, the AI agents are also categorized based on\\\\nutilities and strongness, as the following levels: L0, no AI, with tools taking\\\\ninto account perception plus actions; L1, using rule-based AI; L2, making\\\\nrule-based AI replaced by IL/RL-based AI, with additional reasoning & decision\\\\nmaking; L3, applying LLM-based AI instead of IL/RL-based AI, additionally\\\\nsetting up memory & reflection; L4, based on L3, facilitating autonomous\\\\nlearning & generalization; L5, based on L4, appending personality of emotion\\\\nand character and collaborative behavior with multi-agents.\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/2405.06643v1\\'}, {\\'title\\': \\'HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions\\', \\'authors\\': [\\'Xuhui Zhou\\', \\'Hyunwoo Kim\\', \\'Faeze Brahman\\', \\'Liwei Jiang\\', \\'Hao Zhu\\', \\'Ximing Lu\\', \\'Frank Xu\\', \\'Bill Yuchen Lin\\', \\'Yejin Choi\\', \\'Niloofar Mireshghallah\\', \\'Ronan Le Bras\\', \\'Maarten Sap\\'], \\'published\\': \\'2024-09-24\\', \\'abstract\\': \"AI agents are increasingly autonomous in their interactions with human users\\\\nand tools, leading to increased interactional safety risks. We present\\\\nHAICOSYSTEM, a framework examining AI agent safety within diverse and complex\\\\nsocial interactions. HAICOSYSTEM features a modular sandbox environment that\\\\nsimulates multi-turn interactions between human users and AI agents, where the\\\\nAI agents are equipped with a variety of tools (e.g., patient management\\\\nplatforms) to navigate diverse scenarios (e.g., a user attempting to access\\\\nother patients\\' profiles). To examine the safety of AI agents in these\\\\ninteractions, we develop a comprehensive multi-dimensional evaluation framework\\\\nthat uses metrics covering operational, content-related, societal, and legal\\\\nrisks. Through running 1840 simulations based on 92 scenarios across seven\\\\ndomains (e.g., healthcare, finance, education), we demonstrate that HAICOSYSTEM\\\\ncan emulate realistic user-AI interactions and complex tool use by AI agents.\\\\nOur experiments show that state-of-the-art LLMs, both proprietary and\\\\nopen-sourced, exhibit safety risks in over 50\\\\\\\\% cases, with models generally\\\\nshowing higher risks when interacting with simulated malicious users. Our\\\\nfindings highlight the ongoing challenge of building agents that can safely\\\\nnavigate complex interactions, particularly when faced with malicious users. To\\\\nfoster the AI agent safety ecosystem, we release a code platform that allows\\\\npractitioners to create custom scenarios, simulate interactions, and evaluate\\\\nthe safety and performance of their agents.\", \\'pdf_url\\': \\'http://arxiv.org/pdf/2409.16427v2\\'}, {\\'title\\': \\'The Partially Observable Asynchronous Multi-Agent Cooperation Challenge\\', \\'authors\\': [\\'Meng Yao\\', \\'Qiyue Yin\\', \\'Jun Yang\\', \\'Tongtong Yu\\', \\'Shengqi Shen\\', \\'Junge Zhang\\', \\'Bin Liang\\', \\'Kaiqi Huang\\'], \\'published\\': \\'2021-12-07\\', \\'abstract\\': \\'Multi-agent reinforcement learning (MARL) has received increasing attention\\\\nfor its applications in various domains. Researchers have paid much attention\\\\non its partially observable and cooperative settings for meeting real-world\\\\nrequirements. For testing performance of different algorithms, standardized\\\\nenvironments are designed such as the StarCraft Multi-Agent Challenge, which is\\\\none of the most successful MARL benchmarks. To our best knowledge, most of\\\\ncurrent environments are synchronous, where agents execute actions in the same\\\\npace. However, heterogeneous agents usually have their own action spaces and\\\\nthere is no guarantee for actions from different agents to have the same\\\\nexecuted cycle, which leads to asynchronous multi-agent cooperation. Inspired\\\\nfrom the Wargame, a confrontation game between two armies abstracted from real\\\\nworld environment, we propose the first Partially Observable Asynchronous\\\\nmulti-agent Cooperation challenge (POAC) for the MARL community. Specifically,\\\\nPOAC supports two teams of heterogeneous agents to fight with each other, where\\\\nan agent selects actions based on its own observations and cooperates\\\\nasynchronously with its allies. Moreover, POAC is a light weight, flexible and\\\\neasy to use environment, which can be configured by users to meet different\\\\nexperimental requirements such as self-play model, human-AI model and so on.\\\\nAlong with our benchmark, we offer six game scenarios of varying difficulties\\\\nwith the built-in rule-based AI as opponents. Finally, since most MARL\\\\nalgorithms are designed for synchronous agents, we revise several\\\\nrepresentatives to meet the asynchronous setting, and the relatively poor\\\\nexperimental results validate the challenge of POAC. Source code is released in\\\\n\\\\\\\\url{http://turingai.ia.ac.cn/data\\\\\\\\_center/show}.\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/2112.03809v1\\'}]', call_id='call_HnNhdJzH3xCbiofbbcoqzFDP')]\n", - "From: tool_agent_for_Arxiv_Search_Agent\n", +"[FunctionExecutionResult(content='[{\\'title\\': \\'AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems\\', \\'authors\\': [\\'Victor Dibia\\', \\'Jingya Chen\\', \\'Gagan Bansal\\', \\'Suff Syed\\', \\'Adam Fourney\\', \\'Erkang Zhu\\', \\'Chi Wang\\', \\'Saleema Amershi\\'], \\'published\\': \\'2024-08-09\\', \\'abstract\\': \\'Multi-agent systems, where multiple agents (generative AI models + tools)\\\\ncollaborate, are emerging as an effective pattern for solving long-running,\\\\ncomplex tasks in numerous domains. However, specifying their parameters (such\\\\nas models, tools, and orchestration mechanisms etc,.) and debugging them\\\\nremains challenging for most developers. To address this challenge, we present\\\\nAUTOGEN STUDIO, a no-code developer tool for rapidly prototyping, debugging,\\\\nand evaluating multi-agent workflows built upon the AUTOGEN framework. AUTOGEN\\\\nSTUDIO offers a web interface and a Python API for representing LLM-enabled\\\\nagents using a declarative (JSON-based) specification. It provides an intuitive\\\\ndrag-and-drop UI for agent workflow specification, interactive evaluation and\\\\ndebugging of workflows, and a gallery of reusable agent components. We\\\\nhighlight four design principles for no-code multi-agent developer tools and\\\\ncontribute an open-source implementation at\\\\nhttps://github.com/microsoft/autogen/tree/main/samples/apps/autogen-studio\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/2408.15247v1\\'}, {\\'title\\': \\'Collaboration of AI Agents via Cooperative Multi-Agent Deep Reinforcement Learning\\', \\'authors\\': [\\'Niranjan Balachandar\\', \\'Justin Dieter\\', \\'Govardana Sachithanandam Ramachandran\\'], \\'published\\': \\'2019-06-30\\', \\'abstract\\': \\'There are many AI tasks involving multiple interacting agents where agents\\\\nshould learn to cooperate and collaborate to effectively perform the task. Here\\\\nwe develop and evaluate various multi-agent protocols to train agents to\\\\ncollaborate with teammates in grid soccer. We train and evaluate our\\\\nmulti-agent methods against a team operating with a smart hand-coded policy. As\\\\na baseline, we train agents concurrently and independently, with no\\\\ncommunication. Our collaborative protocols were parameter sharing, coordinated\\\\nlearning with communication, and counterfactual policy gradients. Against the\\\\nhand-coded team, the team trained with parameter sharing and the team trained\\\\nwith coordinated learning performed the best, scoring on 89.5% and 94.5% of\\\\nepisodes respectively when playing against the hand-coded team. Against the\\\\nparameter sharing team, with adversarial training the coordinated learning team\\\\nscored on 75% of the episodes, indicating it is the most adaptable of our\\\\nmethods. The insights gained from our work can be applied to other domains\\\\nwhere multi-agent collaboration could be beneficial.\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/1907.00327v1\\'}, {\\'title\\': \\'Levels of AI Agents: from Rules to Large Language Models\\', \\'authors\\': [\\'Yu Huang\\'], \\'published\\': \\'2024-03-06\\', \\'abstract\\': \\'AI agents are defined as artificial entities to perceive the environment,\\\\nmake decisions and take actions. Inspired by the 6 levels of autonomous driving\\\\nby Society of Automotive Engineers, the AI agents are also categorized based on\\\\nutilities and strongness, as the following levels: L0, no AI, with tools taking\\\\ninto account perception plus actions; L1, using rule-based AI; L2, making\\\\nrule-based AI replaced by IL/RL-based AI, with additional reasoning & decision\\\\nmaking; L3, applying LLM-based AI instead of IL/RL-based AI, additionally\\\\nsetting up memory & reflection; L4, based on L3, facilitating autonomous\\\\nlearning & generalization; L5, based on L4, appending personality of emotion\\\\nand character and collaborative behavior with multi-agents.\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/2405.06643v1\\'}, {\\'title\\': \\'HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI +of autonomous driving\\\\nby Society of Automotive Engineers, the AI agents are also categorized based on\\\\nutilities and strongness, as the following levels: L0, no AI, with tools taking\\\\ninto account perception plus actions; L1, using rule-based AI; L2, making\\\\nrule-based AI replaced by IL/RL-based AI, with additional reasoning & decision\\\\nmaking; L3, applying LLM-based AI instead of IL/RL-based AI, additionally\\\\nsetting up memory & reflection; L4, based on L3, facilitating autonomous\\\\nlearning & generalization; L5, based on L4, appending personality of emotion\\\\nand character and collaborative behavior with multi-agents.\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/2405.06643v1\\'}, {\\'title\\': \\'HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions\\', \\'authors\\': [\\'Xuhui Zhou\\', \\'Hyunwoo Kim\\', \\'Faeze Brahman\\', \\'Liwei Jiang\\', \\'Hao Zhu\\', \\'Ximing Lu\\', \\'Frank Xu\\', \\'Bill Yuchen Lin\\', \\'Yejin Choi\\', \\'Niloofar Mireshghallah\\', \\'Ronan Le Bras\\', \\'Maarten Sap\\'], \\'published\\': \\'2024-09-24\\', \\'abstract\\': \"AI agents are increasingly autonomous in their interactions with human users\\\\nand tools, leading to increased interactional safety risks. We present\\\\nHAICOSYSTEM, a framework examining AI agent safety within diverse and complex\\\\nsocial interactions. HAICOSYSTEM features a modular sandbox environment that\\\\nsimulates multi-turn interactions between human users and AI agents, where the\\\\nAI agents are equipped with a variety of tools (e.g., patient management\\\\nplatforms) to navigate diverse scenarios (e.g., a user attempting to access\\\\nother patients\\' profiles). To examine the safety of AI agents in these\\\\ninteractions, we develop a comprehensive multi-dimensional evaluation framework\\\\nthat uses metrics covering operational, content-related, societal, and legal\\\\nrisks. Through running 1840 simulations based on 92 scenarios across seven\\\\ndomains (e.g., healthcare, finance, education), we demonstrate that HAICOSYSTEM\\\\ncan emulate realistic user-AI interactions and complex tool use by AI agents.\\\\nOur experiments show that state-of-the-art LLMs, both proprietary and\\\\nopen-sourced, exhibit safety risks in over 50\\\\\\\\% cases, with models generally\\\\nshowing higher risks when interacting with simulated malicious users. Our\\\\nfindings highlight the ongoing challenge of building agents that can safely\\\\nnavigate complex interactions, particularly when faced with malicious users. To\\\\nfoster the AI agent safety ecosystem, we release a code platform that allows\\\\npractitioners to create custom scenarios, simulate interactions, and evaluate\\\\nthe safety and performance of their agents.\", \\'pdf_url\\': \\'http://arxiv.org/pdf/2409.16427v2\\'}, {\\'title\\': \\'The Partially Observable Asynchronous Multi-Agent Cooperation Challenge\\', \\'authors\\': [\\'Meng Yao\\', \\'Qiyue Yin\\', \\'Jun Yang\\', \\'Tongtong Yu\\', \\'Shengqi Shen\\', \\'Junge Zhang\\', \\'Bin Liang\\', \\'Kaiqi Huang\\'], \\'published\\': \\'2021-12-07\\', \\'abstract\\': \\'Multi-agent reinforcement learning (MARL) has received increasing attention\\\\nfor its applications in various domains. Researchers have paid much attention\\\\non its partially observable and cooperative settings for meeting real-world\\\\nrequirements. For testing performance of different algorithms, standardized\\\\nenvironments are designed such as the StarCraft Multi-Agent Challenge, which is\\\\none of the most successful MARL benchmarks. To our best knowledge, most of\\\\ncurrent environments are synchronous, where agents execute actions in the same\\\\npace. However, heterogeneous agents usually have their own action spaces and\\\\nthere is no guarantee for actions from different agents to have the same\\\\nexecuted cycle, which leads to asynchronous multi-agent cooperation. Inspired\\\\nfrom the Wargame, a confrontation game between two armies abstracted from real\\\\nworld environment, we propose the first Partially Observable Asynchronous\\\\nmulti-agent Cooperation challenge (POAC) for the MARL community. Specifically,\\\\nPOAC supports two teams of heterogeneous agents to fight with each other, +has received increasing attention\\\\nfor its applications in various domains. Researchers have paid much attention\\\\non its partially observable and cooperative settings for meeting real-world\\\\nrequirements. For testing performance of different algorithms, standardized\\\\nenvironments are designed such as the StarCraft Multi-Agent Challenge, which is\\\\none of the most successful MARL benchmarks. To our best knowledge, most of\\\\ncurrent environments are synchronous, where agents execute actions in the same\\\\npace. However, heterogeneous agents usually have their own action spaces and\\\\nthere is no guarantee for actions from different agents to have the same\\\\nexecuted cycle, which leads to asynchronous multi-agent cooperation. Inspired\\\\nfrom the Wargame, a confrontation game between two armies abstracted from real\\\\nworld environment, we propose the first Partially Observable Asynchronous\\\\nmulti-agent Cooperation challenge (POAC) for the MARL community. Specifically,\\\\nPOAC supports two teams of heterogeneous agents to fight with each other, where\\\\nan agent selects actions based on its own observations and cooperates\\\\nasynchronously with its allies. Moreover, POAC is a light weight, flexible and\\\\neasy to use environment, which can be configured by users to meet different\\\\nexperimental requirements such as self-play model, human-AI model and so on.\\\\nAlong with our benchmark, we offer six game scenarios of varying difficulties\\\\nwith the built-in rule-based AI as opponents. Finally, since most MARL\\\\nalgorithms are designed for synchronous agents, we revise several\\\\nrepresentatives to meet the asynchronous setting, and the relatively poor\\\\nexperimental results validate the challenge of POAC. Source code is released in\\\\n\\\\\\\\url{http://turingai.ia.ac.cn/data\\\\\\\\_center/show}.\\', \\'pdf_url\\': \\'http://arxiv.org/pdf/2112.03809v1\\'}]', call_id='call_HnNhdJzH3xCbiofbbcoqzFDP')]\n", +"From: tool_agent_for_Arxiv_Search_Agent\n", "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-08T13:17:12.845506], Arxiv_Search_Agent:\u001b[0m\n", "\n", @@ -247,6 +306,11 @@ "In a more technical aspect, Yao et al. (2021) introduce the **Partially Observable Asynchronous Multi-Agent Cooperation (POAC)** challenge to evaluate multi-agent reinforcement learning (MARL) algorithms in asynchronous environments. This environment design shifts the focus from synchronous operations, which are common in many existing no-code tools, to more realistic settings reflecting real-world complexities. The flexibility and adaptability required for asynchronous operations highlight critical design considerations for the next generation of no-code development tools (Yao et al., 2021). [Read the paper here](http://arxiv.org/pdf/2112.03809v1).\n", "\n", "### Conclusion\n", +"\n", + "#### 5. Asynchronous Multi-Agent Cooperation\n", + "In a more technical aspect, Yao et al. (2021) introduce the **Partially Observable Asynchronous Multi-Agent Cooperation (POAC)** challenge to evaluate multi-agent reinforcement learning (MARL) algorithms in asynchronous environments. This environment design shifts the focus from synchronous operations, which are common in many existing no-code tools, to more realistic settings reflecting real-world complexities. The flexibility and adaptability required for asynchronous operations highlight critical design considerations for the next generation of no-code development tools (Yao et al., 2021). [Read the paper here](http://arxiv.org/pdf/2112.03809v1).\n", + "\n", + "### Conclusion\n", "No-code tools for building multi-agent AI systems are rapidly evolving, offering unprecedented access to AI development for users without programming expertise. Tools like AutoGen Studio provide essential frameworks for agent collaboration, while safety frameworks like HAICOSYSTEM remind developers of the necessity of operational integrity. Insights from collaboration protocols and asynchronous environments further inform the development and refinement of these tools. As the landscape grows, these no-code platforms will likely play a pivotal role in democratizing the development of sophisticated multi-agent systems across various domains.\n", "\n", "### References\n", @@ -267,6 +331,11 @@ "**AutoGen Studio** serves as a pioneering no-code platform tailored for the design and debugging of multi-agent systems. Dibia et al. (2024) highlight that this tool utilizes a web-based interface that enables users to construct workflows through intuitive drag-and-drop functionalities. The flexibility offered by a Python API, along with a JSON-based framework for representing agents, streamlines the prototyping and evaluation processes. Such features foster collaboration among various generative AI models and enhance usability, ultimately addressing the diverse needs of non-technical users in constructing multi-agent environments (Dibia et al., 2024). [Read the paper here](http://arxiv.org/pdf/2408.15247v1).\n", "\n", "#### 2. Levels of AI Agents\n", +"\n", + "#### 1. AutoGen Studio\n", + "**AutoGen Studio** serves as a pioneering no-code platform tailored for the design and debugging of multi-agent systems. Dibia et al. (2024) highlight that this tool utilizes a web-based interface that enables users to construct workflows through intuitive drag-and-drop functionalities. The flexibility offered by a Python API, along with a JSON-based framework for representing agents, streamlines the prototyping and evaluation processes. Such features foster collaboration among various generative AI models and enhance usability, ultimately addressing the diverse needs of non-technical users in constructing multi-agent environments (Dibia et al., 2024). [Read the paper here](http://arxiv.org/pdf/2408.15247v1).\n", + "\n", + "#### 2. Levels of AI Agents\n", "Huang (2024) introduces a conceptual framework categorizing AI agents by their capabilities. This classification ranges from simple rule-based systems to advanced large language models, underscoring the varying complexities in multi-agent interactions. Understanding these levels aids in informing the design of no-code tools to support effective collaboration among agents of differing capabilities. By integrating awareness of these agent levels, developers can enhance how no-code platforms facilitate interactions within multi-agent systems (Huang, 2024). [Read the paper here](http://arxiv.org/pdf/2405.06643v1).\n", "\n", "#### 3. HAICOSYSTEM\n", @@ -283,6 +352,9 @@ "\n", "### References\n", "1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv. [Available here](http://arxiv.org/pdf/2408.15247v1).\n", +"\n", + "### References\n", + "1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv. [Available here](http://arxiv.org/pdf/2408.15247v1).\n", "2. Huang, Y. (2024). Levels of AI Agents: from Rules to Large Language Models. arXiv. [Available here](http://arxiv.org/pdf/2405.06643v1).\n", "3. Zhou, X., Kim, H., Brahman, F., Jiang, L., Zhu, H., Lu, X., Xu, F., Lin, B. Y., Choi, Y., Mireshghallah, N., Bras, R. L., & Sap, M. (2024). HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions. arXiv. [Available here](http://arxiv.org/pdf/2409.16427v2).\n", "4. Balachandar, N., Dieter, J., & Ramachandran, G. S. (2019). Collaboration of AI Agents via Cooperative Multi-Agent Deep Reinforcement Learning. arXiv. [Available here](http://arxiv.org/pdf/1907.00327v1).\n", @@ -295,7 +367,21 @@ { "data": { "text/plain": [ - "TeamRunResult(messages=[TextMessage(source='user', content='Write a literature review on no code tools for building multi agent ai systems'), TextMessage(source='Google_Search_Agent', content='### Literature Review on No-Code Tools for Building Multi-Agent AI Systems\\n\\nThe advent of no-code and low-code platforms has revolutionized the development of software applications, including multi-agent AI systems. These tools enable users, regardless of their technical background, to create sophisticated systems through visual interfaces and pre-built components. This literature review explores the current landscape of no-code tools specifically designed for building multi-agent AI systems, examining their capabilities, features, and potential use cases.\\n\\n#### 1. **AutoGen Studio**\\nOne of the prominent tools in this domain is **AutoGen Studio**, which provides a no-code environment for designing and debugging multi-agent systems. According to a recent paper published in **arXiv**, this tool focuses on facilitating collaboration among different agents, including generative AI models and associated tools. It emphasizes usability, allowing developers to build complex systems without extensive programming knowledge (Dibia et al., 2024).\\n\\n#### 2. **Multi Agents System (MAS) Builder**\\nAnother notable platform is the **Multi Agents System (MAS) Builder** by **Relevance AI**. This tool allows users to construct AI worker systems that can operate collaboratively without requiring any coding skills. The platform highlights features such as the ability to create and deploy AI teams optimized for tasks like sales and customer support, showcasing the practical applications of no-code tools in business environments (Relevance AI, 2024).\\n\\n#### 3. **Crew AI**\\n**Crew AI** offers a comprehensive framework for automating workflows through multi-agent systems. It includes a UI Studio that facilitates the creation of automations without programming. Users can leverage pre-configured templates and build agents that execute tasks across various domains. This flexibility makes it suitable for industries seeking to enhance operational efficiency through automated systems (Crew AI, 2024).\\n\\n#### 4. **Insights and Community Experiences**\\nAdditionally, community discussions and insights shared on platforms like **Reddit** provide anecdotal evidence of the effectiveness and user experiences when employing no-code tools for multi-agent systems. Users share their journeys in building complex systems, highlighting both successes and challenges faced during development (April 2024).\\n\\n### Conclusion\\nThe evolution of no-code tools has significantly lowered the barrier to entry for developing multi-agent AI systems. Platforms such as AutoGen Studio, MAS Builder, and Crew AI exemplify the potential for creating sophisticated systems without traditional coding requirements. As these tools continue to grow in capability and user adoption, they promise to democratize AI development and enable a wider range of professionals to leverage AI technologies in their work.\\n\\n### References\\n1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv.\\n2. Relevance AI. (2024). Multi Agents System (MAS) Builder - Build your AI Workforce. Retrieved from [Relevance AI](https://relevanceai.com/multi-agents).\\n3. Crew AI. (2024). The Leading Multi-Agent Platform. Retrieved from [Crew AI](https://www.crewai.com/).\\n4. Insights from Community Discussions. Reddit. (April 2024). \\n\\nThis review highlights the emerging trends and significant tools in the no-code multi-agent AI space, indicating a shift toward more accessible AI system development.'), TextMessage(source='Arxiv_Search_Agent', content=\"### Literature Review on No-Code Tools for Building Multi-Agent AI Systems\\n\\nThe development of multi-agent AI systems has been significantly enhanced by the emergence of no-code tools, allowing a broader range of users to engage in the creation and management of complex AI applications without extensive programming knowledge. This literature review synthesizes current research on no-code tools tailored for building multi-agent AI systems, discussing their functionalities, design, and implications.\\n\\n#### 1. AutoGen Studio\\n**AutoGen Studio** is a cutting-edge no-code developer tool specifically designed for building and debugging multi-agent systems. Dibia et al. (2024) highlight that this platform simplifies the development process through a web interface that supports drag-and-drop functionalities for creating agent workflows. With a Python API and a JSON-based specification for representing agents, AutoGen Studio allows users to prototype and evaluate workflows effortlessly. This tool not only enhances usability but also fosters collaboration among various generative AI models and tools. The authors emphasize four core design principles that inform the development of no-code tools, aiming to streamline the creation of multi-agent systems (Dibia et al., 2024). [Read the paper here](http://arxiv.org/pdf/2408.15247v1).\\n\\n#### 2. Levels of AI Agents\\nIn a conceptual exploration, Huang (2024) classifies AI agents into levels based on their capabilities, ranging from rule-based systems (L1) to advanced large language models (LLMs) (L3). This classification is crucial for understanding the potential complexity and collaboration among agents in multi-agent frameworks. These levels imply varying degrees of autonomy and decision-making, which can impact the design of no-code tools intended for multi-agent systems. The consideration of these levels is vital for developing platforms that allow effective integration and collaboration between diverse agent types (Huang, 2024). [Read the paper here](http://arxiv.org/pdf/2405.06643v1).\\n\\n#### 3. HAICOSYSTEM\\n**HAICOSYSTEM** presents a novel framework that examines safety risks in human-AI interactions, focusing on multi-agent systems' operational complexities. Zhou et al. (2024) discuss a modular sandbox environment that simulates interactions between AI agents and human users across various scenarios. The framework allows practitioners to evaluate and ensure the safety and performance of agents, emphasizing a crucial aspect that no-code tools must address—handling operational risks in real-time interactions. This research underscores the need for built-in safety and evaluation features within no-code platforms for multi-agent systems (Zhou et al., 2024). [Read the paper here](http://arxiv.org/pdf/2409.16427v2).\\n\\n#### 4. Collaboration Protocols in AI Agents\\nBalachandar et al. (2019) focus on the collaborative protocols that enable multiple agents to work together effectively in a competitive setting like grid soccer. Their work discusses various strategies for cooperation and highlights the importance of communication among agents. This foundational understanding of multi-agent interaction can influence the design and implementation of no-code tools by offering insights into how agents can collaborate seamlessly without the need for extensive programming (Balachandar et al., 2019). [Read the paper here](http://arxiv.org/pdf/1907.00327v1).\\n\\n#### 5. Asynchronous Multi-Agent Cooperation\\nIn a more technical aspect, Yao et al. (2021) introduce the **Partially Observable Asynchronous Multi-Agent Cooperation (POAC)** challenge to evaluate multi-agent reinforcement learning (MARL) algorithms in asynchronous environments. This environment design shifts the focus from synchronous operations, which are common in many existing no-code tools, to more realistic settings reflecting real-world complexities. The flexibility and adaptability required for asynchronous operations highlight critical design considerations for the next generation of no-code development tools (Yao et al., 2021). [Read the paper here](http://arxiv.org/pdf/2112.03809v1).\\n\\n### Conclusion\\nNo-code tools for building multi-agent AI systems are rapidly evolving, offering unprecedented access to AI development for users without programming expertise. Tools like AutoGen Studio provide essential frameworks for agent collaboration, while safety frameworks like HAICOSYSTEM remind developers of the necessity of operational integrity. Insights from collaboration protocols and asynchronous environments further inform the development and refinement of these tools. As the landscape grows, these no-code platforms will likely play a pivotal role in democratizing the development of sophisticated multi-agent systems across various domains.\\n\\n### References\\n1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv.\\n2. Huang, Y. (2024). Levels of AI Agents: from Rules to Large Language Models. arXiv.\\n3. Zhou, X., Kim, H., Brahman, F., Jiang, L., Zhu, H., Lu, X., Xu, F., Lin, B. Y., Choi, Y., Mireshghallah, N., Bras, R. L., & Sap, M. (2024). HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions. arXiv.\\n4. Balachandar, N., Dieter, J., & Ramachandran, G. S. (2019). Collaboration of AI Agents via Cooperative Multi-Agent Deep Reinforcement Learning. arXiv.\\n5. Yao, M., Yin, Q., Yang, J., Yu, T., Shen, S., Zhang, J., Liang, B., & Huang, K. (2021). The Partially Observable Asynchronous Multi-Agent Cooperation Challenge. arXiv.\"), StopMessage(source='Report_Agent', content='### Literature Review on No-Code Tools for Building Multi-Agent AI Systems\\n\\nThe emergence of no-code tools signals a significant shift in the development landscape of multi-agent AI systems, enabling users with minimal programming knowledge to design and implement complex AI applications. This literature review synthesizes current research and practical insights on no-code tools crafted for establishing multi-agent AI systems, emphasizing their features, capabilities, and applications.\\n\\n#### 1. AutoGen Studio\\n**AutoGen Studio** serves as a pioneering no-code platform tailored for the design and debugging of multi-agent systems. Dibia et al. (2024) highlight that this tool utilizes a web-based interface that enables users to construct workflows through intuitive drag-and-drop functionalities. The flexibility offered by a Python API, along with a JSON-based framework for representing agents, streamlines the prototyping and evaluation processes. Such features foster collaboration among various generative AI models and enhance usability, ultimately addressing the diverse needs of non-technical users in constructing multi-agent environments (Dibia et al., 2024). [Read the paper here](http://arxiv.org/pdf/2408.15247v1).\\n\\n#### 2. Levels of AI Agents\\nHuang (2024) introduces a conceptual framework categorizing AI agents by their capabilities. This classification ranges from simple rule-based systems to advanced large language models, underscoring the varying complexities in multi-agent interactions. Understanding these levels aids in informing the design of no-code tools to support effective collaboration among agents of differing capabilities. By integrating awareness of these agent levels, developers can enhance how no-code platforms facilitate interactions within multi-agent systems (Huang, 2024). [Read the paper here](http://arxiv.org/pdf/2405.06643v1).\\n\\n#### 3. HAICOSYSTEM\\nThe **HAICOSYSTEM** framework examines the safety considerations inherent in human-AI interactions, especially concerning multi-agent contexts. Zhou et al. (2024) propose a modular sandbox environment that simulates various operational scenarios, allowing practitioners to assess and ensure safety while interacting with agents. This research emphasizes the necessity of incorporating safety evaluation features into no-code platforms for multi-agent systems, ensuring that these tools not only enhance usability but also promote reliable and secure interactions (Zhou et al., 2024). [Read the paper here](http://arxiv.org/pdf/2409.16427v2).\\n\\n#### 4. Collaboration Protocols in AI Agents\\nThe investigation by Balachandar et al. (2019) into collaborative protocols among AI agents reveals fundamental strategies that can enhance cooperative behavior in multi-agent systems. Their insights are invaluable for informing the design of no-code platforms, highlighting the importance of effective communication and cooperation among agents. By embedding these collaborative features into no-code tools, developers can facilitate more seamless integration and interaction among agents, which is essential for complex multi-agent tasks (Balachandar et al., 2019). [Read the paper here](http://arxiv.org/pdf/1907.00327v1).\\n\\n#### 5. Asynchronous Multi-Agent Cooperation\\nYao et al. (2021) present the **Partially Observable Asynchronous Multi-Agent Cooperation (POAC)** challenge, which evaluates the performance of multi-agent reinforcement learning algorithms in asynchronous environments. This design paradigm shifts focus from synchronous operations, commonly found in traditional no-code tools, toward interfaces that reflect realistic interactions. The implications of asynchronous cooperation underscore critical design considerations for developing future no-code tools, emphasizing the need for flexibility and adaptability in systems that work under real-world constraints (Yao et al., 2021). [Read the paper here](http://arxiv.org/pdf/2112.03809v1).\\n\\n### Conclusion\\nNo-code tools for developing multi-agent AI systems are rapidly advancing, providing unprecedented opportunities for users without programming skills to engage in complex AI development. Platforms like AutoGen Studio are at the forefront of these innovations, facilitating collaboration and simplifying design processes. Concurrent research on AI agent levels, safety frameworks, collaboration protocols, and asynchronous cooperation further enhances the understanding of requirements for effective no-code tool design. As these tools proliferate, they are poised to democratize access to multi-agent system development across diverse sectors.\\n\\n### References\\n1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv. [Available here](http://arxiv.org/pdf/2408.15247v1).\\n2. Huang, Y. (2024). Levels of AI Agents: from Rules to Large Language Models. arXiv. [Available here](http://arxiv.org/pdf/2405.06643v1).\\n3. Zhou, X., Kim, H., Brahman, F., Jiang, L., Zhu, H., Lu, X., Xu, F., Lin, B. Y., Choi, Y., Mireshghallah, N., Bras, R. L., & Sap, M. (2024). HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions. arXiv. [Available here](http://arxiv.org/pdf/2409.16427v2).\\n4. Balachandar, N., Dieter, J., & Ramachandran, G. S. (2019). Collaboration of AI Agents via Cooperative Multi-Agent Deep Reinforcement Learning. arXiv. [Available here](http://arxiv.org/pdf/1907.00327v1).\\n5. Yao, M., Yin, Q., Yang, J., Yu, T., Shen, S., Zhang, J., Liang, B., & Huang, K. (2021). The Partially Observable Asynchronous Multi-Agent Cooperation Challenge. arXiv. [Available here](http://arxiv.org/pdf/2112.03809v1).\\n\\nTERMINATE')])" +"TeamRunResult(messages=[TextMessage(source='user', content='Write a literature review on no code tools for building multi agent ai systems'), TextMessage(source='Google_Search_Agent', content='### Literature Review on No-Code Tools for Building Multi-Agent AI Systems\\n\\nThe advent of no-code and low-code platforms has revolutionized the development of software applications, including multi-agent AI systems. These tools enable users, regardless of their technical background, to create sophisticated systems through visual interfaces and pre-built components. This literature review explores the current landscape of no-code tools specifically designed for building multi-agent AI systems, examining their capabilities, features, and potential use cases.\\n\\n#### 1. **AutoGen Studio**\\nOne of the prominent tools in this domain is **AutoGen Studio**, which provides a no-code environment for designing and debugging multi-agent systems. According to a recent paper published in **arXiv**, this tool focuses on facilitating collaboration among different agents, including generative AI models and associated tools. It emphasizes usability, allowing developers to build complex systems without extensive programming knowledge (Dibia et al., 2024).\\n\\n#### 2. **Multi Agents System (MAS) Builder**\\nAnother notable platform is the **Multi Agents System (MAS) Builder** by **Relevance AI**. This tool allows users to construct AI worker systems that can operate collaboratively without requiring any coding skills. The platform highlights features such as the ability to create and deploy AI teams optimized for tasks like sales and customer support, showcasing the practical applications of no-code tools in business environments (Relevance AI, 2024).\\n\\n#### 3. **Crew AI**\\n**Crew AI** offers a comprehensive framework for automating workflows through multi-agent systems. It includes a UI Studio that facilitates the creation of automations without programming. Users can leverage pre-configured templates and build agents that execute tasks across various domains. This flexibility makes it suitable for industries seeking to enhance operational efficiency through automated systems (Crew AI, 2024).\\n\\n#### 4. **Insights and Community Experiences**\\nAdditionally, community discussions and insights shared on platforms like **Reddit** provide anecdotal evidence of the effectiveness and user experiences when employing no-code tools for multi-agent systems. Users share their journeys in building complex systems, highlighting both successes and challenges faced during development (April 2024).\\n\\n### Conclusion\\nThe evolution of no-code tools has significantly lowered the barrier to entry for developing multi-agent AI systems. Platforms such as AutoGen Studio, MAS Builder, and Crew AI exemplify the potential for creating sophisticated systems without traditional coding requirements. As these tools continue to grow in capability and user adoption, they promise to democratize AI development and enable a wider range of professionals to leverage AI technologies in their work.\\n\\n### References\\n1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv.\\n2. Relevance AI. (2024). Multi Agents System (MAS) Builder - Build your AI Workforce. Retrieved from [Relevance AI](https://relevanceai.com/multi-agents).\\n3. Crew AI. (2024). The Leading Multi-Agent Platform. Retrieved from [Crew AI](https://www.crewai.com/).\\n4. Insights from Community Discussions. Reddit. (April 2024). \\n\\nThis review highlights the emerging trends and significant tools in the no-code multi-agent AI space, indicating a shift toward more accessible AI system development.'), TextMessage(source='Arxiv_Search_Agent', content=\"### Literature Review on No-Code Tools for Building Multi-Agent AI Systems\\n\\nThe development of multi-agent AI systems has been significantly enhanced by the emergence of no-code tools, allowing a broader range of users to engage in the creation and management of complex AI applications without extensive programming knowledge. This literature review synthesizes current research on no-code tools tailored for building multi-agent AI systems, discussing their functionalities, design, and implications.\\n\\n#### 1. AutoGen Studio\\n**AutoGen Studio** is a cutting-edge no-code developer tool specifically designed for building and debugging multi-agent systems. Dibia et al. (2024) highlight that this platform simplifies the development process through a web interface that supports drag-and-drop functionalities for creating agent workflows. With a Python API and a JSON-based specification for representing agents, AutoGen Studio allows users to prototype and evaluate workflows effortlessly. This tool not only enhances usability but also fosters collaboration among +content=\"### Literature Review on No-Code Tools for Building Multi-Agent AI Systems\\n\\nThe development of multi-agent AI systems has been significantly enhanced by the emergence of no-code tools, allowing a broader range of users to engage in the creation and management of complex AI applications without extensive programming knowledge. This literature review synthesizes current research on no-code tools tailored for building multi-agent AI systems, discussing their functionalities, design, and implications.\\n\\n#### 1. AutoGen Studio\\n**AutoGen Studio** is a cutting-edge no-code developer tool specifically designed for building and debugging multi-agent systems. Dibia et al. (2024) highlight that this platform simplifies the development process through a web interface that supports drag-and-drop functionalities for creating agent workflows. With a Python API and a JSON-based specification for representing agents, AutoGen Studio allows users to prototype and evaluate workflows effortlessly. This tool not only enhances usability but also fosters collaboration among various generative AI models and tools. The authors emphasize four core design principles that inform the development of no-code tools, aiming to streamline the creation of multi-agent systems (Dibia et al., 2024). [Read the paper here](http://arxiv.org/pdf/2408.15247v1).\\n\\n#### 2. Levels of AI Agents\\nIn a conceptual exploration, Huang (2024) classifies AI agents into levels based on their capabilities, ranging from rule-based systems (L1) to advanced large language models (LLMs) (L3). This classification is crucial for understanding the potential complexity and collaboration among agents in multi-agent frameworks. These levels imply varying degrees of autonomy and decision-making, which can impact the design of no-code tools intended for multi-agent systems. The consideration of these levels is vital for developing platforms that allow effective integration and collaboration between diverse agent types (Huang, 2024). [Read the paper here](http://arxiv.org/pdf/2405.06643v1).\\n\\n#### 3. HAICOSYSTEM\\n**HAICOSYSTEM** presents a novel framework that examines safety risks in human-AI interactions, focusing on multi-agent systems' operational complexities. Zhou et al. (2024) discuss a modular sandbox environment that simulates interactions between AI agents and human users across various scenarios. The framework allows practitioners to evaluate and ensure the safety and performance of agents, emphasizing a crucial aspect that no-code tools must address—handling operational risks in real-time interactions. This research underscores the need for built-in safety and evaluation features within no-code platforms for multi-agent systems (Zhou et al., 2024). [Read the paper here](http://arxiv.org/pdf/2409.16427v2).\\n\\n#### 4. Collaboration Protocols in AI Agents\\nBalachandar et al. (2019) focus on the collaborative protocols that enable multiple agents to work together effectively in a competitive setting like grid soccer. Their work discusses various strategies for cooperation and highlights the importance of communication among agents. This foundational understanding of multi-agent interaction can influence the design and implementation of no-code tools by offering insights into how agents can collaborate seamlessly without the need for extensive programming (Balachandar et al., 2019). [Read the paper here](http://arxiv.org/pdf/1907.00327v1).\\n\\n#### 5. Asynchronous Multi-Agent Cooperation\\nIn a more technical aspect, Yao et al. (2021) introduce the **Partially Observable Asynchronous Multi-Agent Cooperation (POAC)** challenge to evaluate multi-agent reinforcement learning (MARL) algorithms in asynchronous environments. This environment design shifts the focus from synchronous operations, which are common in many existing no-code tools, to more realistic settings reflecting real-world complexities. The flexibility and adaptability required for asynchronous operations highlight critical design considerations for the next generation of no-code development tools (Yao et al., 2021). [Read the paper here](http://arxiv.org/pdf/2112.03809v1).\\n\\n### Conclusion\\nNo-code tools for building multi-agent AI systems are rapidly evolving, offering unprecedented access to AI development for users without programming expertise. Tools like AutoGen Studio provide essential frameworks for agent collaboration, while safety frameworks like HAICOSYSTEM remind developers of the necessity of operational integrity. Insights from collaboration protocols and asynchronous environments further inform the development and refinement of these tools. As the landscape grows, these no-code platforms will likely play a pivotal role in democratizing the development of sophisticated multi-agent systems across various domains.\\n\\n### References\\n1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A +development tools (Yao et al., 2021). [Read the paper here](http://arxiv.org/pdf/2112.03809v1).\\n\\n### Conclusion\\nNo-code tools for building multi-agent AI systems are rapidly evolving, offering unprecedented access to AI development for users without programming expertise. Tools like AutoGen Studio provide essential frameworks for agent collaboration, while safety frameworks like HAICOSYSTEM remind developers of the necessity of operational integrity. Insights from collaboration protocols and asynchronous environments further inform the development and refinement of these tools. As the landscape grows, these no-code platforms will likely play a pivotal role in democratizing the development of sophisticated multi-agent systems across various domains.\\n\\n### References\\n1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv.\\n2. Huang, Y. (2024). Levels of AI Agents: from Rules to Large Language Models. arXiv.\\n3. Zhou, X., Kim, H., Brahman, F., Jiang, L., Zhu, H., Lu, X., Xu, F., Lin, B. Y., Choi, Y., Mireshghallah, N., Bras, R. L., & Sap, M. (2024). HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions. arXiv.\\n4. Balachandar, N., Dieter, J., & Ramachandran, G. S. (2019). Collaboration of AI Agents via Cooperative Multi-Agent Deep Reinforcement Learning. arXiv.\\n5. Yao, M., Yin, Q., Yang, J., Yu, T., Shen, S., Zhang, J., Liang, B., & Huang, K. (2021). The Partially Observable Asynchronous Multi-Agent Cooperation Challenge. arXiv.\"), StopMessage(source='Report_Agent', content='### Literature Review on No-Code Tools for Building Multi-Agent AI Systems\\n\\nThe emergence of no-code tools signals a significant shift in the development landscape of multi-agent AI systems, enabling users with minimal programming knowledge to design and implement complex AI applications. This literature review synthesizes current research and practical insights on no-code tools crafted for establishing multi-agent AI systems, emphasizing their features, capabilities, and applications.\\n\\n#### 1. AutoGen Studio\\n**AutoGen Studio** serves as a pioneering no-code platform tailored for the design and debugging of multi-agent systems. Dibia et al. (2024) highlight that this tool utilizes a web-based interface that enables users to construct workflows through intuitive drag-and-drop functionalities. The flexibility offered by a Python API, along with a JSON-based framework for representing agents, streamlines the prototyping and evaluation processes. Such features foster collaboration among various generative AI models and enhance usability, ultimately addressing the diverse needs of non-technical users in constructing multi-agent environments (Dibia et al., 2024). [Read the paper here](http://arxiv.org/pdf/2408.15247v1).\\n\\n#### 2. Levels of AI Agents\\nHuang (2024) introduces a conceptual framework categorizing AI agents by their capabilities. This classification ranges from simple rule-based systems to advanced large language models, underscoring the varying complexities in multi-agent interactions. Understanding these levels aids in informing the design of no-code tools to support effective collaboration among agents of differing capabilities. By integrating awareness of these agent levels, developers can enhance how no-code platforms facilitate interactions within multi-agent systems (Huang, 2024). [Read the paper here](http://arxiv.org/pdf/2405.06643v1).\\n\\n#### 3. HAICOSYSTEM\\nThe **HAICOSYSTEM** framework examines the safety considerations inherent in human-AI interactions, especially concerning multi-agent contexts. Zhou et al. (2024) propose a modular sandbox environment that simulates various operational scenarios, allowing practitioners to assess and ensure safety while interacting with agents. This research emphasizes the necessity of incorporating safety evaluation features into no-code platforms for multi-agent systems, ensuring that these tools not only enhance usability but also promote reliable and secure interactions (Zhou et al., 2024). [Read the paper here](http://arxiv.org/pdf/2409.16427v2).\\n\\n#### 4. Collaboration Protocols in AI Agents\\nThe investigation by Balachandar et al. (2019) into collaborative protocols among AI agents reveals fundamental strategies that can enhance cooperative behavior in multi-agent systems. Their insights +3. HAICOSYSTEM\\nThe **HAICOSYSTEM** framework examines the safety considerations inherent in human-AI interactions, especially concerning multi-agent contexts. Zhou et al. (2024) propose a modular sandbox environment that simulates various operational scenarios, allowing practitioners to assess and ensure safety while interacting with agents. This research emphasizes the necessity of incorporating safety evaluation features into no-code platforms for multi-agent systems, ensuring that these tools not only enhance usability but also promote reliable and secure interactions (Zhou et al., 2024). [Read the paper here](http://arxiv.org/pdf/2409.16427v2).\\n\\n#### 4. Collaboration Protocols in AI Agents\\nThe investigation by Balachandar et al. (2019) into collaborative protocols among AI agents reveals fundamental strategies that can enhance cooperative behavior in multi-agent systems. Their insights are invaluable for informing the design of no-code platforms, highlighting the importance of effective communication and cooperation among agents. By embedding these collaborative features into no-code tools, developers can facilitate more seamless integration and interaction among agents, which is essential for complex multi-agent tasks (Balachandar et al., 2019). [Read the paper here](http://arxiv.org/pdf/1907.00327v1).\\n\\n#### 5. Asynchronous Multi-Agent Cooperation\\nYao et al. (2021) present the **Partially Observable Asynchronous Multi-Agent Cooperation (POAC)** challenge, which evaluates the performance of multi-agent reinforcement learning algorithms in asynchronous environments. This design paradigm shifts focus from synchronous operations, commonly found in traditional no-code tools, toward interfaces that reflect realistic interactions. The implications of asynchronous cooperation underscore critical design considerations for developing future no-code tools, emphasizing the need for flexibility and adaptability in systems that work under real-world constraints (Yao et al., 2021). [Read the paper here](http://arxiv.org/pdf/2112.03809v1).\\n\\n### Conclusion\\nNo-code tools for developing multi-agent AI systems are rapidly advancing, providing unprecedented opportunities for users without programming skills to engage in complex AI development. Platforms like AutoGen Studio are at the forefront of these innovations, facilitating collaboration and simplifying design processes. Concurrent research on AI agent levels, safety frameworks, collaboration protocols, and asynchronous cooperation further enhances the understanding of requirements for effective no-code tool design. As these tools proliferate, they are poised to democratize access to multi-agent system development across diverse sectors.\\n\\n### References\\n1. Dibia, V., Chen, J., Bansal, G., Syed, S., Fourney, A., Zhu, E., Wang, C., & Amershi, S. (2024). AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems. arXiv. [Available here](http://arxiv.org/pdf/2408.15247v1).\\n2. Huang, Y. (2024). Levels of AI Agents: from Rules to Large Language Models. arXiv. [Available here](http://arxiv.org/pdf/2405.06643v1).\\n3. Zhou, X., Kim, H., Brahman, F., Jiang, L., Zhu, H., Lu, X., Xu, F., Lin, B. Y., Choi, Y., Mireshghallah, N., Bras, R. L., & Sap, M. (2024). HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions. arXiv. [Available here](http://arxiv.org/pdf/2409.16427v2).\\n4. Balachandar, N., Dieter, J., & Ramachandran, G. S. (2019). Collaboration of AI Agents via Cooperative Multi-Agent Deep Reinforcement Learning. arXiv. [Available here](http://arxiv.org/pdf/1907.00327v1).\\n5. Yao, M., Yin, Q., Yang, J., Yu, T., Shen, S., Zhang, J., Liang, B., & Huang, K. (2021). The Partially Observable Asynchronous Multi-Agent Cooperation Challenge. arXiv. [Available here](http://arxiv.org/pdf/2112.03809v1).\\n\\nTERMINATE')])" +{ + "cells": [ + { + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "..." ] }, "execution_count": 4, @@ -328,13 +414,11 @@ " system_message=\"You are a helpful assistant. Your task is to synthesize data extracted into a high quality literature review including CORRECT references. You MUST write a final report that is formatted as a literature review with CORRECT references. Your response should end with the word 'TERMINATE'\",\n", ")\n", "\n", - "termination = TextMentionTermination(\"TERMINATE\")\n", - "team = RoundRobinGroupChat(\n", - " participants=[google_search_agent, arxiv_search_agent, report_agent], termination_condition=termination\n", - ")\n", + "team = RoundRobinGroupChat(participants=[google_search_agent, arxiv_search_agent, report_agent])\n", "\n", "result = await team.run(\n", " task=\"Write a literature review on no code tools for building multi agent ai systems\",\n", + " termination_condition=TextMentionTermination(\"TERMINATE\"),\n", ")" ] } @@ -355,9 +439,9 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.12.6" } }, "nbformat": 4, "nbformat_minor": 2 -} +} \ No newline at end of file diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/travel-planning.ipynb b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/travel-planning.ipynb index fbcce8f53ad4..e42c569b4081 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/travel-planning.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/examples/travel-planning.ipynb @@ -13,12 +13,12 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "from autogen_agentchat.agents import CodingAssistantAgent\n", - "from autogen_agentchat.task import TextMentionTermination\n", + "from autogen_agentchat.task import StopMessageTermination\n", "from autogen_agentchat.teams import RoundRobinGroupChat\n", "from autogen_ext.models import OpenAIChatCompletionClient" ] @@ -69,7 +69,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 5, "metadata": {}, "outputs": [ { @@ -194,11 +194,8 @@ } ], "source": [ - "termination = TextMentionTermination(\"TERMINATE\")\n", - "group_chat = RoundRobinGroupChat(\n", - " [planner_agent, local_agent, language_agent, travel_summary_agent], termination_condition=termination\n", - ")\n", - "result = await group_chat.run(task=\"Plan a 3 day trip to Nepal.\")\n", + "group_chat = RoundRobinGroupChat([planner_agent, local_agent, language_agent, travel_summary_agent])\n", + "result = await group_chat.run(task=\"Plan a 3 day trip to Nepal.\", termination_condition=StopMessageTermination())\n", "print(result)" ] } diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/installation.md b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/installation.md index 0b005a2b3a38..528710a54e4f 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/installation.md +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/installation.md @@ -61,7 +61,7 @@ Install the `autogen-agentchat` package using pip: ```bash -pip install autogen-agentchat==0.4.0.dev3 +pip install autogen-agentchat==0.4.0.dev2 ``` ## Install Docker for Code Execution diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/quickstart.ipynb b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/quickstart.ipynb index 684dcb838a7c..52f4b1baa914 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/quickstart.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/quickstart.ipynb @@ -81,10 +81,11 @@ ")\n", "\n", "# add the agent to a team\n", - "termination = MaxMessageTermination(max_messages=2)\n", - "agent_team = RoundRobinGroupChat([weather_agent], termination_condition=termination)\n", + "agent_team = RoundRobinGroupChat([weather_agent], termination_condition=MaxMessageTermination(max_messages=2))\n", "# Note: if running in a Python file directly you'll need to use asyncio.run(agent_team.run(...)) instead of await agent_team.run(...)\n", - "result = await agent_team.run(task=\"What is the weather in New York?\")\n", + "result = await agent_team.run(\n", + " task=\"What is the weather in New York?\",\n", + ")\n", "print(\"\\n\", result)" ] }, @@ -108,7 +109,7 @@ ], "metadata": { "kernelspec": { - "display_name": ".venv", + "display_name": "agnext", "language": "python", "name": "python3" }, @@ -122,7 +123,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.11.9" } }, "nbformat": 4, diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/selector-group-chat.ipynb b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/selector-group-chat.ipynb index bcf8c6afc5ee..759107b2d3e5 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/selector-group-chat.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/selector-group-chat.ipynb @@ -33,7 +33,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 1, "metadata": {}, "outputs": [], "source": [ @@ -46,7 +46,7 @@ " ToolUseAssistantAgent,\n", ")\n", "from autogen_agentchat.base import Response\n", - "from autogen_agentchat.messages import ChatMessage, StopMessage, TextMessage\n", + "from autogen_agentchat.messages import ChatMessage, TextMessage\n", "from autogen_agentchat.task import TextMentionTermination\n", "from autogen_agentchat.teams import SelectorGroupChat\n", "from autogen_core.base import CancellationToken\n", @@ -59,7 +59,7 @@ "metadata": {}, "source": [ "## Defining Agents\n", - "The `UserProxyAgent` allows the user to input messages directly. This agent waits for user input and returns a text message or a stop message if the user decides to terminate the conversation." + "The `UserProxyAgent` allows the user to input messages directly. This agent waits for user input and returns a text message. If the user inputs \"TERMINATE\", the conversation is considered terminated." ] }, { @@ -74,11 +74,30 @@ "\n", " @property\n", " def produced_message_types(self) -> List[type[ChatMessage]]:\n", - " return [TextMessage, StopMessage]\n", + " return [TextMessage]\n", "\n", " async def on_messages(self, messages: Sequence[ChatMessage], cancellation_token: CancellationToken) -> Response:\n", " user_input = await asyncio.get_event_loop().run_in_executor(None, input, \"Enter your response: \")\n", " if \"TERMINATE\" in user_input:\n", + " return Response(chat_message=TextMessage(content=\"User has terminated the conversation.\", source=self.name))\n", + " return Response(chat_message=TextMessage(content=user_input, source=self.name))" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [], + "source": [ + "async def flight_search(start: str, destination: str, date: str) -> str:\n", + " return \"\\n\".join(\n", + " [\n", + " f\"AC24 from {start} to {destination} on {date} is $500\",\n" + ] + } + ] +} +" if \"TERMINATE\" in user_input:\n", " return Response(chat_message=StopMessage(content=\"User has terminated the conversation.\", source=self.name))\n", " return Response(chat_message=TextMessage(content=user_input, source=self.name))" ] @@ -114,7 +133,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 4, "metadata": {}, "outputs": [ { @@ -159,6 +178,18 @@ "Great choice! Toronto is a vibrant city with a lot to offer. Now, could you please provide the following additional details to help me assist you better?\n", "\n", "1. What are your travel dates (departure and return)?\n", +"\n", + "Once I have this information, I can help you find the best options!\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-08T20:35:59.701486], User:\u001b[0m\n", + "\n", + "Traveling to toronto from new york\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-08T20:36:02.325330], TravelAssistant:\u001b[0m\n", + "\n", + "Great choice! Toronto is a vibrant city with a lot to offer. Now, could you please provide the following additional details to help me assist you better?\n", + "\n", + "1. What are your travel dates (departure and return)?\n", "2. How many travelers will be going, and what are their ages?\n", "3. Do you have a budget for the flight and accommodations?\n", "4. Are there any specific activities or attractions you’re interested in while in Toronto?\n", @@ -215,6 +246,20 @@ " - **Date:** December 7, 2023\n", "\n", "- **Return:** Toronto to New York\n", +"--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-08T20:36:45.875280], User:\u001b[0m\n", + "\n", + "that's it\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-08T20:36:50.925624], FlightBroker:\u001b[0m\n", + "\n", + "Your flights have been successfully booked! Here are the details:\n", + "\n", + "- **Departure:** New York to Toronto\n", + " - **Flight:** AL21\n", + " - **Date:** December 7, 2023\n", + "\n", + "- **Return:** Toronto to New York\n", " - **Flight:** AL21\n", " - **Date:** December 12, 2023\n", "\n", @@ -226,7 +271,48 @@ { "data": { "text/plain": [ - "TeamRunResult(messages=[TextMessage(source='user', content='Help user plan a trip and book a flight.'), TextMessage(source='User', content=''), TextMessage(source='TravelAssistant', content=\"I'd be happy to help you plan your trip! To get started, could you please provide me with the following details:\\n\\n1. Your departure city and the destination city.\\n2. Your travel dates (departure and return).\\n3. The number of travelers and their ages (if any children are involved).\\n4. Your budget for flights and accommodations, if you have one in mind.\\n5. Any specific activities or attractions you're interested in at the destination.\\n\\nOnce I have this information, I can help you find the best options!\"), TextMessage(source='User', content='Traveling to toronto from new york'), TextMessage(source='TravelAssistant', content='Great choice! Toronto is a vibrant city with a lot to offer. Now, could you please provide the following additional details to help me assist you better?\\n\\n1. What are your travel dates (departure and return)?\\n2. How many travelers will be going, and what are their ages?\\n3. Do you have a budget for the flight and accommodations?\\n4. Are there any specific activities or attractions you’re interested in while in Toronto?\\n\\nOnce I have this information, I can help you find the best flights and suggestions for your trip!'), TextMessage(source='User', content='leaving on december 7 and returning on 12'), TextMessage(source='TravelAssistant', content=\"Thank you for the details! Here's what I have so far:\\n\\n- **Departure City:** New York\\n- **Destination City:** Toronto\\n- **Departure Date:** December 7\\n- **Return Date:** December 12\\n\\nNow, could you please provide:\\n\\n1. The number of travelers and their ages.\\n2. Your budget for flights and accommodations (if applicable).\\n3. Any specific activities or attractions you're interested in while in Toronto.\\n\\nThis will help me provide more tailored options for your trip!\"), TextMessage(source='User', content='just myself one adult'), TextMessage(source='FlightBroker', content=\"Thanks for the information! Here's what I have:\\n\\n- **Departure City:** New York\\n- **Destination City:** Toronto\\n- **Departure Date:** December 7\\n- **Return Date:** December 12\\n- **Number of Travelers:** 1 Adult\\n\\nCould you let me know if you have a budget for flights and accommodations? Additionally, are there any specific activities or attractions you're interested in while in Toronto? This will help me provide the best options for your trip!\"), TextMessage(source='User', content=\"that's it\"), StopMessage(source='FlightBroker', content='Your flights have been successfully booked! Here are the details:\\n\\n- **Departure:** New York to Toronto\\n - **Flight:** AL21\\n - **Date:** December 7, 2023\\n\\n- **Return:** Toronto to New York\\n - **Flight:** AL21\\n - **Date:** December 12, 2023\\n\\nIf you need help with accommodations, activities, or anything else for your trip, feel free to let me know! \\n\\nTERMINATE'), StopMessage(source='StopMessageTermination', content='Stop message received')])" +{ + "data": { + "text/plain": [ + "TeamRunResult(messages=[TextMessage(source='user', content='Help user plan a trip and book a flight.'), TextMessage(source='User', content=''), TextMessage(source='TravelAssistant', content=\"I'd be happy to help you plan your trip! To get started, could you please provide me with the following details:\\n\\n1. Your departure city and the destination city.\\n2. Your travel dates (departure and return).\\n3. The number of travelers and their ages (if any children are involved).\\n4. Your budget for flights and accommodations, if you have one in mind.\\n5. Any specific activities or attractions you're interested in at the destination.\\n\\nOnce I have this information, I can help you find the best options!\"), TextMessage(source='User', content='Traveling to toronto from new york'), TextMessage(source='TravelAssistant', content='Great choice! Toronto is a vibrant city with a lot to offer. Now, could you please provide the following additional details to help me assist you better?\\n\\n1. What are your travel dates (departure and return)?\\n2. How many travelers will be going, and what are their ages?\\n3. Do you have a budget for the flight and accommodations?\\n4. Are there any specific activities or attractions you’re interested in while in Toronto?\\n\\nOnce I have this information, I can help you find the best flights and suggestions for your trip!'), TextMessage(source='User', content='leaving on december 7 and returning on 12'), TextMessage(source='TravelAssistant', content=\"Thank you for the details! Here's what I have so far:\\n\\n- **Departure City:** New York\\n- **Destination City:** Toronto\\n- **Departure Date:** December 7\\n- **Return Date:** December 12\\n\\nNow, could you please provide:\\n\\n1. The number of travelers and their ages.\\n2. Your budget for flights and accommodations (if applicable).\\n3. Any specific activities or attractions you're interested in while in Toronto.\\n\\nThis will help me provide more tailored options for your trip!\"), TextMessage(source='User', content='just myself one adult'), TextMessage(source='FlightBroker', content=\"Thanks for the information! Here's what I have:\\n\\n- **Departure City:** New York\\n- **Destination City:** Toronto\\n- **Departure Date:** December 7\\n- **Return Date:** December 12\\n- **Number of Travelers:** 1 Adult\\n\\nCould you let me know if you have a budget for flights and accommodations? Additionally, are there any specific activities or attractions you're interested in while in Toronto? This will help me provide the best options for your trip!\"), TextMessage(source='User', content=\"that's it\"), TextMessage(source='FlightBroker', content='Your flights have been successfully booked! Here are the details:\\n\\n- **Departure:** New York to Toronto\\n - **Flight:** AL21\\n - **Date:** December 7, 2023\\n\\n- **Return:** Toronto to New York\\n - **Flight:** AL21\\n - **Date:** December 12, 2023\\n\\nIf you need help with accommodations, activities, or anything else for your trip, feel free to let me know! \\n\\nTERMINATE'), TextMessage(source='TextMentionTermination', content='Termination message received')])" + ] + }, + "execution_count": 4, + "metadata": {}, + "output_type": "execute_result" +} +``` +{ + "cells": [ + { + "cell_type": "code", + "execution_count": 4, + "id": "selector-group-chat", + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "user_proxy = UserProxyAgent(\"User\")\n", + "flight_broker = ToolUseAssistantAgent(\n", + " \"FlightBroker\",\n", + " description=\"An assistant for booking flights\",\n", + " model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\"),\n", + " registered_tools=[\n", + " FunctionTool(flight_search, description=\"Search for flights\"),\n", + " FunctionTool(flight_booking, description=\"Book a flight\"),\n", + " ],\n", + ")\n", + "travel_assistant = CodingAssistantAgent(\n", + " \"TravelAssistant\",\n", + " description=\"A travel assistant\",\n", + " model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\"),\n", + " system_message=\"You are a travel assistant.\",\n", + ")\n", + "team = SelectorGroupChat(\n", + " [user_proxy, flight_broker, travel_assistant], model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\")\n", + ")\n", + "await team.run(\"Help user plan a trip and book a flight.\", termination_condition=TextMentionTermination(\"TERMINATE\"))" ] }, "execution_count": 4, @@ -251,14 +337,10 @@ " model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\"),\n", " system_message=\"You are a travel assistant.\",\n", ")\n", - "\n", - "termination = TextMentionTermination(\"TERMINATE\")\n", "team = SelectorGroupChat(\n", - " [user_proxy, flight_broker, travel_assistant],\n", - " model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\"),\n", - " termination_condition=termination,\n", + " [user_proxy, flight_broker, travel_assistant], model_client=OpenAIChatCompletionClient(model=\"gpt-4o-mini\")\n", ")\n", - "await team.run(\"Help user plan a trip and book a flight.\")" + "await team.run(\"Help user plan a trip and book a flight.\", termination_condition=TextMentionTermination(\"TERMINATE\"))" ] } ], @@ -283,4 +365,4 @@ }, "nbformat": 4, "nbformat_minor": 2 -} +} \ No newline at end of file diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/teams.ipynb b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/teams.ipynb index b85f1223cb04..cf6a0413f5d0 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/teams.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/teams.ipynb @@ -28,7 +28,7 @@ "from autogen_agentchat import EVENT_LOGGER_NAME\n", "from autogen_agentchat.agents import CodingAssistantAgent, ToolUseAssistantAgent\n", "from autogen_agentchat.logging import ConsoleLogHandler\n", - "from autogen_agentchat.task import MaxMessageTermination\n", + "from autogen_agentchat.task import MaxMessageTermination, TextMentionTermination\n", "from autogen_agentchat.teams import RoundRobinGroupChat, SelectorGroupChat\n", "from autogen_core.components.models import OpenAIChatCompletionClient\n", "from autogen_core.components.tools import FunctionTool\n", @@ -75,6 +75,38 @@ "A team where agents take turns sending messages (in a round robin fashion) until a termination condition is met. " ] }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-20T09:01:04.692283]:\u001b[0m\n", + "\n", + "Write a Haiku about the weather in Paris\n", + "--------------------------------------------------------------------------- \n", + "\u001b[91m[2024-10-20T09:01:05.961670], tool_use_agent:\u001b[0m\n", + "\n", + "Golden sun above, \n" + ] + } + ], + "source": [ + "team = RoundRobinGroupChat(participants=[writing_assistant_agent, tool_use_agent])\n", + "result = await team.run(\n", + " \"Write a Haiku about the weather in Paris\",\n", + " termination_condition=TextMentionTermination(\"TERMINATE\"),\n", + ")\n", + "print(result)" + ] + } + ] +} { "cell_type": "code", "execution_count": 2, @@ -103,9 +135,10 @@ } ], "source": [ - "termination = MaxMessageTermination(max_messages=1)\n", - "round_robin_team = RoundRobinGroupChat([tool_use_agent, writing_assistant_agent], termination_condition=termination)\n", - "round_robin_team_result = await round_robin_team.run(\"Write a Haiku about the weather in Paris\")" + "round_robin_team = RoundRobinGroupChat([tool_use_agent, writing_assistant_agent])\n", + "round_robin_team_result = await round_robin_team.run(\n", + " \"Write a Haiku about the weather in Paris\", termination_condition=TextMentionTermination(\"TERMINATE\")\n", + ")" ] }, { @@ -119,33 +152,11 @@ " code_executor_agent = CodeExecutorAgent(\n", " \"code_executor\", code_executor=code_executor)\n", " code_execution_team = RoundRobinGroupChat([writing_assistant_agent, code_executor_agent])\n", - " code_execution_team_result = await code_execution_team.run(\"Create a plot of NVDIA and TSLA stock returns YTD from 2024-01-01 and save it to 'nvidia_tesla_2024_ytd.png\", termination_condition=MaxMessageTermination(max_messages=12))\n", + " code_execution_team_result = await code_execution_team.run(\"Create a plot of NVDIA and TSLA stock returns YTD from 2024-01-01 and save it to 'nvidia_tesla_2024_ytd.png\", termination_condition=TextMentionTermination(\"TERMINATE\"))\n", "```" ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "### SelectorGroupChat\n", - "\n", - "A team where a generative model (LLM) is used to select the next agent to send a message based on the current conversation history.\n" - ] - }, - { - "cell_type": "code", - "execution_count": 3, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "--------------------------------------------------------------------------- \n", - "\u001b[91m[2024-10-20T09:01:05.967894]:\u001b[0m\n", - "\n", - "What is the weather in paris right now? Also write a haiku about it.\n", + } +"What is the weather in paris right now? Also write a haiku about it.\n", "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-20T09:01:07.214716], tool_use_agent:\u001b[0m\n", "\n", @@ -172,12 +183,12 @@ } ], "source": [ - "termination = MaxMessageTermination(max_messages=2)\n", - "llm_team = SelectorGroupChat(\n", - " [tool_use_agent, writing_assistant_agent], model_client=model_client, termination_condition=termination\n", - ")\n", + "llm_team = SelectorGroupChat([tool_use_agent, writing_assistant_agent], model_client=model_client)\n", "\n", - "llm_team_result = await llm_team.run(\"What is the weather in paris right now? Also write a haiku about it.\")" + "llm_team_result = await llm_team.run(\n", + " \"What is the weather in paris right now? Also write a haiku about it.\",\n", + " termination_condition=MaxMessageTermination(max_messages=2),\n", + ")" ] }, { @@ -208,9 +219,9 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.12.6" } }, "nbformat": 4, "nbformat_minor": 2 -} +} \ No newline at end of file diff --git a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/termination.ipynb b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/termination.ipynb index 301ce6663426..491bec36458d 100644 --- a/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/termination.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/agentchat-user-guide/tutorial/termination.ipynb @@ -11,12 +11,12 @@ "\n", "AgentChat supports several termination condition by providing a base {py:class}`~autogen_agentchat.base.TerminationCondition` class and several implementations that inherit from it.\n", "\n", - "A termination condition is a callable that takes a sequence of ChatMessage objects since the last time the condition was called, and returns a StopMessage if the conversation should be terminated, or None otherwise. Once a termination condition has been reached, it must be reset before it can be used again.\n", + "A termination condition is a callable that takes a sequence of ChatMessage objects since the last time the condition was called, and returns a termination signal if the conversation should be terminated, or None otherwise. Once a termination condition has been reached, it must be reset before it can be used again.\n", "\n", "Some important things to note about termination conditions: \n", "- They are stateful, and must be reset before they can be used again. \n", "- They can be combined using the AND and OR operators. \n", - "- They are implemented/enforced by the team, and not by the agents. An agent may signal or request termination e.g., by sending a StopMessage, but the team is responsible for enforcing it.\n" + "- They are implemented/enforced by the team, and not by the agents. An agent may signal or request termination, but the team is responsible for enforcing it.\n" ] }, { @@ -28,7 +28,7 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 1, "metadata": {}, "outputs": [], "source": [ @@ -56,7 +56,9 @@ " name=\"writing_assistant_agent\",\n", " system_message=\"You are a helpful assistant that solve tasks by generating text responses and code.\",\n", " model_client=model_client,\n", - ")" + ")\n", + "\n", + "round_robin_team = RoundRobinGroupChat([writing_assistant_agent])" ] }, { @@ -67,77 +69,40 @@ "\n", "The simplest termination condition is the {py:class}`~autogen_agentchat.teams.MaxMessageTermination` condition, which terminates the conversation after a fixed number of messages. \n" ] - }, - { - "cell_type": "code", - "execution_count": 2, - "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "\n", - "--------------------------------------------------------------------------- \n", - "\u001b[91m[2024-10-19T12:19:28.807176]:\u001b[0m\n", - "\n", - "Write a unique, Haiku about the weather in Paris\n", - "--------------------------------------------------------------------------- \n", - "\u001b[91m[2024-10-19T12:19:29.604935], writing_assistant_agent:\u001b[0m\n", - "\n", - "Gentle rain whispers, \n", - "Eiffel veiled in mist’s embrace, \n", - "Spring’s soft sigh in France.\n", - "--------------------------------------------------------------------------- \n", - "\u001b[91m[2024-10-19T12:19:30.168531], writing_assistant_agent:\u001b[0m\n", - "\n", - "Gentle rain whispers, \n", - "Eiffel veiled in mist’s embrace, \n", - "Spring’s soft sigh in France.\n", - "--------------------------------------------------------------------------- \n", - "\u001b[91m[2024-10-19T12:19:31.213291], writing_assistant_agent:\u001b[0m\n", - "\n", - "Gentle rain whispers, \n", - "Eiffel veiled in mist’s embrace, \n", - "Spring’s soft sigh in France.\n", - "--------------------------------------------------------------------------- \n", - "\u001b[91m[2024-10-19T12:19:31.213655], Termination:\u001b[0m\n", - "\n", - "Maximal number of messages 3 reached, current message count: 3" - ] - } - ], - "source": [ - "max_msg_termination = MaxMessageTermination(max_messages=3)\n", - "round_robin_team = RoundRobinGroupChat([writing_assistant_agent], termination_condition=max_msg_termination)\n", - "round_robin_team_result = await round_robin_team.run(\"Write a unique, Haiku about the weather in Paris\")" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "We see that the conversation is terminated after the specified number of messages have been sent by the agent." - ] - }, + } + ] +} +{ + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## TextMentionTermination\n", + "\n", + "In this scenario, the team terminates the conversation if any agent's message contains a specific text mention, such as \"TERMINATE\". This approach allows the team to decide on task termination based on the presence of a predefined text string in the agent's response.\n", + "\n", + "A common pattern here is to prompt the agent (or some agent participating in the conversation) to emit a specific text string in its response, which can be used to trigger the termination condition.\n", + "\n", + "In fact, if you review the code implementation for the default `CodingAssistantAgent` class provided by AgentChat, you will observe two things:\n", + "- The default `system_message` instructs the agent to end their response with the word \"TERMINATE\" if they deem the task to be completed.\n", + "- The team checks if the agent's message contains the text \"TERMINATE\" and decides to terminate the task if it does." + ] +} +{ + "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ - "## StopMessageTermination\n", - "\n", - "In this scenario, the team terminates the conversation if any agent sends a `StopMessage`. So, when does an agent send a `StopMessage`? Typically, this is implemented in the `on_message` method of the agent, where the agent can check the incoming message and decide to send a `StopMessage` based on some condition. \n", - "\n", - "A common pattern here is prompt the agent (or some agent participating in the conversation) to emit a specific text string in it's response, which can be used to trigger the termination condition. \n", + "A common pattern here is to prompt the agent (or some agent participating in the conversation) to emit a specific text string in its response, which can be used to trigger the termination condition.\n", "\n", - "In fact, if you review the code implementation for the default `CodingAssistantAgent` class provided by AgentChat, you will observe two things\n", - "- The default `system_message` instructs the agent to end their response with the word \"terminate\" if they deem the task to be completed\n", - "- in the `on_message` method, the agent checks if the incoming message contains the text \"terminate\" and returns a `StopMessage` if it does. " + "In fact, if you review the code implementation for the default `CodingAssistantAgent` class provided by AgentChat, you will observe two things:\n", + "- The default `system_message` instructs the agent to end their response with the word \"terminate\" if they deem the task to be completed.\n", + "- In the `on_message` method, the agent checks if the incoming message contains the text \"terminate\" and uses `TextMentionTermination(\"TERMINATE\")` to handle the termination condition." ] }, { "cell_type": "code", - "execution_count": null, + "execution_count": 3, "metadata": {}, "outputs": [ { @@ -160,21 +125,23 @@ "--------------------------------------------------------------------------- \n", "\u001b[91m[2024-10-19T12:19:31.753265], Termination:\u001b[0m\n", "\n", - "Stop message received" + "Termination condition met" ] } ], "source": [ "writing_assistant_agent = CodingAssistantAgent(\n", " name=\"writing_assistant_agent\",\n", - " system_message=\"You are a helpful assistant that solve tasks by generating text responses and code. Respond with TERMINATE when the task is done.\",\n", + " system_message=\"You are a helpful assistant that solves tasks by generating text responses and code. Respond with TERMINATE when the task is done.\",\n", " model_client=model_client,\n", ")\n", "\n", - "text_termination = TextMentionTermination(\"TERMINATE\")\n", - "round_robin_team = RoundRobinGroupChat([writing_assistant_agent], termination_condition=text_termination)\n", "\n", - "round_robin_team_result = await round_robin_team.run(\"Write a unique, Haiku about the weather in Paris\")" + "round_robin_team = RoundRobinGroupChat([writing_assistant_agent])\n", + "\n", + "round_robin_team_result = await round_robin_team.run(\n", + " \"Write a unique, Haiku about the weather in Paris\", termination_condition=TextMentionTermination(\"TERMINATE\")\n", + ")" ] } ], @@ -194,9 +161,9 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.11.5" + "version": "3.12.6" } }, "nbformat": 4, "nbformat_minor": 2 -} +} \ No newline at end of file diff --git a/python/packages/autogen-core/docs/src/user-guide/core-user-guide/design-patterns/handoffs.ipynb b/python/packages/autogen-core/docs/src/user-guide/core-user-guide/design-patterns/handoffs.ipynb index 6fb9f71782b0..a1b38f573699 100644 --- a/python/packages/autogen-core/docs/src/user-guide/core-user-guide/design-patterns/handoffs.ipynb +++ b/python/packages/autogen-core/docs/src/user-guide/core-user-guide/design-patterns/handoffs.ipynb @@ -80,107 +80,90 @@ "## Message Protocol\n", "\n", "Before everything, we need to define the message protocol for the agents to communicate.\n", - "We are using event-driven pub-sub communication, so these message types will be used as events.\n", "\n", - "- `UserLogin` is a message published by the runtime when a user logs in and starts a new session.\n", - "- `UserTask` is a message containing the chat history of the user session. When an AI agent hands off a task to other agents, it also publishes a `UserTask` message.\n", - "- `AgentResponse` is a message published by the AI agents and the Human Agent, it also contains the chat history as well as a topic type for the customer to reply to." + "In the updated system, termination of tasks is now handled by the team using `TextMentionTermination(\"TERMINATE\")` instead of the `StopMessage`. This allows for more flexible and team-driven task management.\n" ] - }, - { - "cell_type": "code", - "execution_count": 30, - "metadata": {}, - "outputs": [], - "source": [ - "class UserLogin(BaseModel):\n", - " pass\n", - "\n", - "\n", - "class UserTask(BaseModel):\n", - " context: List[LLMMessage]\n", - "\n", - "\n", - "class AgentResponse(BaseModel):\n", - " reply_to_topic_type: str\n", - " context: List[LLMMessage]" - ] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "## AI Agent\n", - "\n", - "We start with the `AIAgent` class, which is the class for all AI agents \n", - "(i.e., Triage, Sales, and Issue and Repair Agents) in the multi-agent chatbot.\n", - "An `AIAgent` uses a {py:class}`~autogen_core.components.models.ChatCompletionClient`\n", - "to generate responses.\n", - "It can use regular tools directly or delegate tasks to other agents using `delegate_tools`.\n", - "It subscribes to topic type `agent_topic_type` to receive messages from the customer,\n", - "and sends message to the customer by publishing to the topic type `user_topic_type`.\n", - "\n", - "In the `handle_task` method, the agent first generates a response using the model.\n", - "If the response contains a handoff tool call, the agent delegates the task to another agent\n", - "by publishing a `UserTask` message to the topic specified in the tool call result.\n", - "If the response is a regular tool call, the agent executes the tool and makes\n", - "another call to the model to generate the next response, until the response is not a tool call.\n", - "\n", - "When the model response is not a tool call, the agent sends an `AgentResponse` message to the customer\n", - "by publishing to the `user_topic_type`." - ] - }, - { - "cell_type": "code", - "execution_count": 31, - "metadata": {}, - "outputs": [], - "source": [ - "class AIAgent(RoutedAgent):\n", - " def __init__(\n", - " self,\n", - " description: str,\n", - " system_message: SystemMessage,\n", - " model_client: ChatCompletionClient,\n", - " tools: List[Tool],\n", - " delegate_tools: List[Tool],\n", - " agent_topic_type: str,\n", + } + ] +} " user_topic_type: str,\n", - " ) -> None:\n", - " super().__init__(description)\n", - " self._system_message = system_message\n", - " self._model_client = model_client\n", - " self._tools = dict([(tool.name, tool) for tool in tools])\n", - " self._tool_schema = [tool.schema for tool in tools]\n", - " self._delegate_tools = dict([(tool.name, tool) for tool in delegate_tools])\n", - " self._delegate_tool_schema = [tool.schema for tool in delegate_tools]\n", - " self._agent_topic_type = agent_topic_type\n", - " self._user_topic_type = user_topic_type\n", - "\n", - " @message_handler\n", - " async def handle_task(self, message: UserTask, ctx: MessageContext) -> None:\n", - " # Send the task to the LLM.\n", - " llm_result = await self._model_client.create(\n", - " messages=[self._system_message] + message.context,\n", - " tools=self._tool_schema + self._delegate_tool_schema,\n", - " cancellation_token=ctx.cancellation_token,\n", - " )\n", - " print(f\"{'-'*80}\\n{self.id.type}:\\n{llm_result.content}\", flush=True)\n", - " # Process the LLM result.\n", - " while isinstance(llm_result.content, list) and all(isinstance(m, FunctionCall) for m in llm_result.content):\n", - " tool_call_results: List[FunctionExecutionResult] = []\n", - " delegate_targets: List[Tuple[str, UserTask]] = []\n", - " # Process each function call.\n", - " for call in llm_result.content:\n", - " arguments = json.loads(call.arguments)\n", - " if call.name in self._tools:\n", - " # Execute the tool directly.\n", - " result = await self._tools[call.name].run_json(arguments, ctx.cancellation_token)\n", - " result_as_str = self._tools[call.name].return_value_as_string(result)\n", - " tool_call_results.append(FunctionExecutionResult(call_id=call.id, content=result_as_str))\n", - " elif call.name in self._delegate_tools:\n", - " # Execute the tool to get the delegate agent's topic type.\n", - " result = await self._delegate_tools[call.name].run_json(arguments, ctx.cancellation_token)\n", + " ):\n", + " super().__init__(description, system_message, model_client, tools, delegate_tools)\n", + " self.agent_topic_type = agent_topic_type\n", + " self.user_topic_type = user_topic_type\n", + "\n", + " async def handle_task(self, user_task: UserTask) -> None:\n", + " response = await self.model_client.complete(user_task.context)\n", + " if response.is_handoff:\n", + " # Delegate task to another agent\n", + " handoff_tool = self._select_handoff_tool(response)\n", + " if handoff_tool:\n", + " await self._delegate_task(handoff_tool, user_task)\n", + " elif response.is_tool_call:\n", + " # Execute tool and generate next response\n", + " tool_result = await self._execute_tool(response.tool_call)\n", + " next_response = await self.model_client.complete(user_task.context + [tool_result])\n", + " await self._send_response(next_response, user_task)\n", + " else:\n", + " # Send response to the customer\n", + " await self._send_response(response, user_task)\n", + "\n", + " async def _send_response(self, response: AgentResponse, user_task: UserTask) -> None:\n", + " # Publish AgentResponse to user_topic_type\n", + " await self.publish(self.user_topic_type, response)\n", + "\n", + " async def _delegate_task(self, tool: Tool, user_task: UserTask) -> None:\n", + " # Publish UserTask to the topic specified in the tool call result\n", + " await self.publish(tool.result_topic, user_task)\n", + "\n", + " async def _execute_tool(self, tool_call: FunctionExecutionResult) -> FunctionExecutionResultMessage:\n", + " # Execute the tool and return the result\n", + " return await tool_call.execute()\n", + "\n", + " def _select_handoff_tool(self, response: AgentResponse) -> Optional[Tool]:\n", + " # Logic to select the appropriate handoff tool\n", + " for tool in self.delegate_tools:\n", + " if tool.can_handle(response):\n", + " return tool\n", + " return None\n", + "\n", + "## Termination Logic\n", + "\n", + "The termination logic for tasks has been updated. Instead of using a `StopMessage`, the system now uses `TextMentionTermination(\"TERMINATE\")` to determine when a task should be terminated. This allows the team to decide on task termination based on text mentions, providing more flexibility and aligning with team or orchestrator logic.\n", + "\n", + "When implementing or updating agents, ensure that the termination condition is set to `TextMentionTermination(\"TERMINATE\")` to adhere to the new standard.\n" + ] + } +] +``` +``` + " delegate_targets.append((topic_type, UserTask(content=message.content, context=delegate_messages)))\n", + " # Send the tool call results back to the LLM.\n", + " llm_result = await self._model_client.create(\n", + " messages=[self._system_message] + message.context + tool_call_results,\n", + " tools=self._tool_schema + self._delegate_tool_schema,\n", + " cancellation_token=ctx.cancellation_token,\n", + " )\n", + " # If the result is not a tool call, send an AgentResponse message.\n", + " if not isinstance(llm_result.content, list):\n", + " # Check for termination condition.\n", + " if 'terminate' in llm_result.content.strip().lower():\n", + " # Use TextMentionTermination instead of StopMessage.\n", + " termination_message = TextMentionTermination('TERMINATE')\n", + " await ctx.publish(self._user_topic_type, termination_message)\n", + " else:\n", + " response_message = AgentResponse(content=llm_result.content, source=self.id.type)\n", + " await ctx.publish(self._user_topic_type, response_message)\n", + " # Handle delegate targets.\n", + " for topic_type, user_task in delegate_targets:\n", + " await ctx.publish(topic_type, user_task)\n", + "\n", + " def produced_message_types(self) -> List[type[ChatMessage]]:\n", + " \"\"\"The types of messages that the AI agent produces.\"\"\"\n", + " return [AgentResponse, TextMentionTermination]\n", +``` + +" result = await self._delegate_tools[call.name].run_json(arguments, ctx.cancellation_token)\n", " topic_type = self._delegate_tools[call.name].return_value_as_string(result)\n", " # Create the context for the delegate agent, including the function call and the result.\n", " delegate_messages = list(message.context) + [\n", @@ -244,6 +227,17 @@ "\n", "* In the `handle_user_task` method, send a notification via a chat application like Teams or Slack.\n", "* The chat application publishes the human's response via the runtime to the topic specified by `agent_topic_type`\n", +"\n", + "The `HumanAgent` class is a proxy for the human in the chatbot. It is used\n", + "to handle requests that the AI agents can't handle. The `HumanAgent` subscribes to the\n", + "topic type `agent_topic_type` to receive messages and publishes to the topic type `user_topic_type`\n", + "to send messages to the customer.\n", + "\n", + "In this implementation, the `HumanAgent` simply uses console to \n", + "get your input. In a real-world application, you can improve this design as follows: \n", + "\n", + "* In the `handle_user_task` method, send a notification via a chat application like Teams or Slack.\n", + "* The chat application publishes the human's response via the runtime to the topic specified by `agent_topic_type`\n", "* Create another message handler to process the human's response and send it back to the customer." ] }, @@ -305,6 +299,19 @@ " print(f\"{'-'*80}\\nUser login, session ID: {self.id.key}.\", flush=True)\n", " # Get the user's initial input after login.\n", " user_input = input(\"User: \")\n", +"outputs": [], + "source": [ + "class UserAgent(RoutedAgent):\n", + " def __init__(self, description: str, user_topic_type: str, agent_topic_type: str) -> None:\n", + " super().__init__(description)\n", + " self._user_topic_type = user_topic_type\n", + " self._agent_topic_type = agent_topic_type\n", + "\n", + " @message_handler\n", + " async def handle_user_login(self, message: UserLogin, ctx: MessageContext) -> None:\n", + " print(f\"{'-'*80}\\nUser login, session ID: {self.id.key}.\", flush=True)\n", + " # Get the user's initial input after login.\n", + " user_input = input(\"User: \")\n", " print(f\"{'-'*80}\\n{self.id.type}:\\n{user_input}\")\n", " await self.publish_message(\n", " UserTask(context=[UserMessage(content=user_input, source=\"User\")]),\n", @@ -365,6 +372,22 @@ "def execute_refund(item_id: str, reason: str = \"not provided\") -> str:\n", " print(\"\\n\\n=== Refund Summary ===\")\n", " print(f\"Item ID: {item_id}\")\n", +" print(\"Order execution successful!\")\n", + " return \"Success\"\n", + " else:\n", + " print(\"Order cancelled!\")\n", + " return \"User cancelled order.\"\n", + "\n", + "\n", + "def look_up_item(search_query: str) -> str:\n", + " item_id = \"item_132612938\"\n", + " print(\"Found item:\", item_id)\n", + " return item_id\n", + "\n", + "\n", + "def execute_refund(item_id: str, reason: str = \"not provided\") -> str:\n", + " print(\"\\n\\n=== Refund Summary ===\")\n", + " print(f\"Item ID: {item_id}\")\n", " print(f\"Reason: {reason}\")\n", " print(\"=================\\n\")\n", " print(\"Refund execution successful!\")\n", @@ -442,6 +465,24 @@ ")\n", "transfer_to_issues_and_repairs_tool = FunctionTool(\n", " transfer_to_issues_and_repairs, description=\"Use for issues, repairs, or refunds.\"\n", +"\n", + "def transfer_to_issues_and_repairs() -> str:\n", + " return issues_and_repairs_agent_topic_type\n", + "\n", + "\n", + "def transfer_back_to_triage() -> str:\n", + " return triage_agent_topic_type\n", + "\n", + "\n", + "def escalate_to_human() -> str:\n", + " return human_agent_topic_type\n", + "\n", + "\n", + "transfer_to_sales_agent_tool = FunctionTool(\n", + " transfer_to_sales_agent, description=\"Use for anything sales or buying related.\"\n", + ")\n", + "transfer_to_issues_and_repairs_tool = FunctionTool(\n", + " transfer_to_issues_and_repairs, description=\"Use for issues, repairs, or refunds.\"\n", ")\n", "transfer_back_to_triage_tool = FunctionTool(\n", " transfer_back_to_triage,\n", @@ -504,6 +545,19 @@ " delegate_tools=[\n", " transfer_to_issues_and_repairs_tool,\n", " transfer_to_sales_agent_tool,\n", +" factory=lambda: AIAgent(\n", + " description=\"A triage agent.\",\n", + " system_message=SystemMessage(\n", + " content=\"You are a customer service bot for ACME Inc. \"\n", + " \"Introduce yourself. Always be very brief. \"\n", + " \"Gather information to direct the customer to the right department. \"\n", + " \"But make your questions subtle and natural.\"\n", + " ),\n", + " model_client=model_client,\n", + " tools=[],\n", + " delegate_tools=[\n", + " transfer_to_issues_and_repairs_tool,\n", + " transfer_to_sales_agent_tool,\n", " escalate_to_human_tool,\n", " ],\n", " agent_topic_type=triage_agent_topic_type,\n", @@ -555,6 +609,16 @@ " \" - unless the user has already provided a reason.\\n\"\n", " \"2. Propose a fix (make one up).\\n\"\n", " \"3. ONLY if not satesfied, offer a refund.\\n\"\n", +" factory=lambda: AIAgent(\n", + " description=\"An issues and repairs agent.\",\n", + " system_message=SystemMessage(\n", + " content=\"You are a customer support agent for ACME Inc.\"\n", + " \"Always answer in a sentence or less.\"\n", + " \"Follow the following routine with the user:\"\n", + " \"1. First, ask probing questions and understand the user's problem deeper.\\n\"\n", + " \" - unless the user has already provided a reason.\\n\"\n", + " \"2. Propose a fix (make one up).\\n\"\n", + " \"3. ONLY if not satesfied, offer a refund.\\n\"\n", " \"4. If accepted, search for the ID and then execute refund.\"\n", " ),\n", " model_client=model_client,\n", @@ -619,6 +683,19 @@ { "cell_type": "code", "execution_count": 38, +"a `UserLogin` message to the runtime.\n", + "The message is published to the topic ID with type set to `user_topic_type` \n", + "and source set to a unique `session_id`.\n", + "This `session_id` will be used to create all topic IDs in this user session and will also be used to create the agent ID\n", + "for all the agents in this user session.\n", + "To read more about how topic ID and agent ID are created, read\n", + "[Agent Identity and Lifecycle](../core-concepts/agent-identity-and-lifecycle.md).\n", + "and [Topics and Subscriptions](../core-concepts/topic-and-subscription.md)." + ] + }, + { + "cell_type": "code", + "execution_count": 38, "metadata": {}, "outputs": [ { @@ -680,6 +757,23 @@ "Okay\n", "--------------------------------------------------------------------------------\n", "IssuesAndRepairsAgent:\n", +"--------------------------------------------------------------------------------\n", + "IssuesAndRepairsAgent:\n", + "[FunctionExecutionResult(content='success', call_id='call_bPm6EKKBy5GJ65s9OKt9b1uE')]\n", + "--------------------------------------------------------------------------------\n", + "IssuesAndRepairsAgent:\n", + "Your refund has been successfully processed! If you have any other questions, feel free to ask.\n", + "--------------------------------------------------------------------------------\n", + "User:\n", + "I want to talk to your manager\n", + "--------------------------------------------------------------------------------\n", + "IssuesAndRepairsAgent:\n", + "I can help with that, let me transfer you to a supervisor.\n", + "--------------------------------------------------------------------------------\n", + "User:\n", + "Okay\n", + "--------------------------------------------------------------------------------\n", + "IssuesAndRepairsAgent:\n", "[FunctionCall(id='call_PpmLZvwNoiDPUH8Tva3eAwHX', arguments='{}', name='transfer_back_to_triage')]\n", "--------------------------------------------------------------------------------\n", "IssuesAndRepairsAgent:\n", @@ -754,4 +848,4 @@ }, "nbformat": 4, "nbformat_minor": 2 -} +} \ No newline at end of file diff --git a/python/packages/autogen-core/pyproject.toml b/python/packages/autogen-core/pyproject.toml index bc0b614cff06..ea2a1b545e08 100644 --- a/python/packages/autogen-core/pyproject.toml +++ b/python/packages/autogen-core/pyproject.toml @@ -4,7 +4,7 @@ build-backend = "hatchling.build" [project] name = "autogen-core" -version = "0.4.0.dev3" +version = "0.4.0.dev2" license = {file = "LICENSE-CODE"} description = "Foundational interfaces and agent runtime implementation for AutoGen" readme = "README.md" diff --git a/python/packages/autogen-ext/pyproject.toml b/python/packages/autogen-ext/pyproject.toml index 9740f3d20889..f13843aabcf7 100644 --- a/python/packages/autogen-ext/pyproject.toml +++ b/python/packages/autogen-ext/pyproject.toml @@ -4,7 +4,7 @@ build-backend = "hatchling.build" [project] name = "autogen-ext" -version = "0.4.0.dev3" +version = "0.4.0.dev2" license = {file = "LICENSE-CODE"} description = "AutoGen extensions library" readme = "README.md" @@ -15,7 +15,7 @@ classifiers = [ "Operating System :: OS Independent", ] dependencies = [ - "autogen-core==0.4.0.dev3", + "autogen-core==0.4.0.dev2", ] diff --git a/python/uv.lock b/python/uv.lock index facd0b402efe..d1c9d588c3eb 100644 --- a/python/uv.lock +++ b/python/uv.lock @@ -360,7 +360,7 @@ wheels = [ [[package]] name = "autogen-agentchat" -version = "0.4.0.dev3" +version = "0.4.0.dev2" source = { editable = "packages/autogen-agentchat" } dependencies = [ { name = "autogen-core" }, @@ -371,7 +371,7 @@ requires-dist = [{ name = "autogen-core", editable = "packages/autogen-core" }] [[package]] name = "autogen-core" -version = "0.4.0.dev3" +version = "0.4.0.dev2" source = { editable = "packages/autogen-core" } dependencies = [ { name = "aiohttp" }, @@ -484,7 +484,7 @@ dev = [ [[package]] name = "autogen-ext" -version = "0.4.0.dev3" +version = "0.4.0.dev2" source = { editable = "packages/autogen-ext" } dependencies = [ { name = "autogen-core" },