v0.11.x (#15398)

.github/workflows/build_package.yml

@@ -28,8 +28,6 @@ jobs:
         uses: actions/setup-python@v5
         with:
           python-version: ${{ matrix.python-version }}
-          cache: "poetry"
-          cache-dependency-path: "**/poetry.lock"
       - name: Install deps
         shell: bash
         run: poetry install

.gitignore

@@ -15,3 +15,4 @@ credentials.json
 token.json
 .python-version
 .DS_Store
+/storage/

CHANGELOG.md

@@ -1,5 +1,17 @@
 # ChangeLog
 
+## [2024-08-22]
+
+### `llama-index-core` [0.11.0]
+
+- removed deprecated `ServiceContext` -- using this now will print an error with a link to the migration guide
+- removed deprecated `LLMPredictor` -- using this now will print an error, any existing LLM is a drop-in replacement
+- made `pandas` an optional dependency
+
+### `Everything Else`
+
+- bumped the minor version of every package to account for the new version of `llama-index-core`
+
 ## [2024-08-21]
 
 ### `llama-index-core` [0.10.68]

docs/BUILD

@@ -0,0 +1,9 @@
+python_sources()
+
+poetry_requirements(
+    name="poetry",
+)
+
+python_requirements(
+    name="reqs",
+)

llama-index-cli/llama_index/cli/rag/base.py

@@ -7,6 +7,7 @@
 from typing import Any, Callable, Dict, Optional, Union, cast
 
 from llama_index.core import (
+    Settings,
     SimpleDirectoryReader,
     VectorStoreIndex,
 )
@@ -16,9 +17,8 @@
     StreamingResponse,
     Response,
 )
-from llama_index.core.bridge.pydantic import BaseModel, Field, validator
+from llama_index.core.bridge.pydantic import BaseModel, Field, field_validator
 from llama_index.core.chat_engine import CondenseQuestionChatEngine
-from llama_index.core.indices.service_context import ServiceContext
 from llama_index.core.ingestion import IngestionPipeline
 from llama_index.core.llms import LLM
 from llama_index.core.query_engine import CustomQueryEngine
@@ -100,7 +100,7 @@ class RagCLI(BaseModel):
     class Config:
         arbitrary_types_allowed = True
 
-    @validator("query_pipeline", always=True)
+    @field_validator("query_pipeline", mode="before")
     def query_pipeline_from_ingestion_pipeline(
         cls, query_pipeline: Any, values: Dict[str, Any]
     ) -> Optional[QueryPipeline]:
@@ -127,15 +127,13 @@ def query_pipeline_from_ingestion_pipeline(
                     embed_model = transformation
                     break
 
-        service_context = ServiceContext.from_defaults(
-            llm=llm, embed_model=embed_model or "default"
-        )
+        Settings.llm = llm
+        Settings.embed_model = embed_model
+
         retriever = VectorStoreIndex.from_vector_store(
-            ingestion_pipeline.vector_store, service_context=service_context
+            ingestion_pipeline.vector_store,
         ).as_retriever(similarity_top_k=8)
-        response_synthesizer = CompactAndRefine(
-            service_context=service_context, streaming=True, verbose=verbose
-        )
+        response_synthesizer = CompactAndRefine(streaming=True, verbose=verbose)
 
         # define query pipeline
         query_pipeline = QueryPipeline(verbose=verbose)
@@ -151,7 +149,7 @@ def query_pipeline_from_ingestion_pipeline(
         query_pipeline.add_link("query", "summarizer", dest_key="query_str")
         return query_pipeline
 
-    @validator("chat_engine", always=True)
+    @field_validator("chat_engine", mode="before")
     def chat_engine_from_query_pipeline(
         cls, chat_engine: Any, values: Dict[str, Any]
     ) -> Optional[CondenseQuestionChatEngine]:

llama-index-cli/llama_index/cli/upgrade/mappings.json

@@ -1,6 +1,5 @@
 {
   "StorageContext": "llama_index.core",
-  "ServiceContext": "llama_index.core",
   "ComposableGraph": "llama_index.core",
   "# indicesSummaryIndex": "llama_index.core",
   "VectorStoreIndex": "llama_index.core",

llama-index-cli/pyproject.toml

@@ -32,13 +32,13 @@ maintainers = [
 name = "llama-index-cli"
 packages = [{include = "llama_index/"}]
 readme = "README.md"
-version = "0.1.13"
+version = "0.3.0"
 
 [tool.poetry.dependencies]
 python = ">=3.8.1,<4.0"
-llama-index-core = "^0.10.11.post1"
-llama-index-embeddings-openai = "^0.1.1"
-llama-index-llms-openai = "^0.1.1"
+llama-index-core = "^0.11.0"
+llama-index-embeddings-openai = "^0.2.0"
+llama-index-llms-openai = "^0.2.0"
 
 [tool.poetry.group.dev.dependencies]
 black = {extras = ["jupyter"], version = "<=23.9.1,>=23.7.0"}

llama-index-core/llama_index/core/__init__.py

@@ -1,6 +1,6 @@
 """Init file of LlamaIndex."""
 
-__version__ = "0.10.68.post1"
+__version__ = "0.11.0"
 
 import logging
 from logging import NullHandler
@@ -152,8 +152,5 @@
 # NOTE: keep for backwards compatibility
 SQLContextBuilder = SQLDocumentContextBuilder
 
-# global service context for ServiceContext.from_defaults()
-global_service_context: Optional[ServiceContext] = None
-
 # global tokenizer
 global_tokenizer: Optional[Callable[[str], list]] = None

llama-index-core/llama_index/core/agent/custom/pipeline_worker.py

@@ -15,7 +15,7 @@
     TaskStepOutput,
 )
 from llama_index.core.base.query_pipeline.query import QueryComponent
-from llama_index.core.bridge.pydantic import BaseModel, Field
+from llama_index.core.bridge.pydantic import BaseModel, Field, ConfigDict
 from llama_index.core.callbacks import (
     CallbackManager,
     trace_method,
@@ -72,14 +72,12 @@ class QueryPipelineAgentWorker(BaseModel, BaseAgentWorker):
 
     """
 
+    model_config = ConfigDict(arbitrary_types_allowed=True)
     pipeline: QueryPipeline = Field(..., description="Query pipeline")
     callback_manager: CallbackManager = Field(..., exclude=True)
     task_key: str = Field("task", description="Key to store task in state")
     step_state_key: str = Field("step_state", description="Key to store step in state")
 
-    class Config:
-        arbitrary_types_allowed = True
-
     def __init__(
         self,
         pipeline: QueryPipeline,

llama-index-core/llama_index/core/agent/custom/simple.py

@@ -19,7 +19,7 @@
     TaskStep,
     TaskStepOutput,
 )
-from llama_index.core.bridge.pydantic import BaseModel, Field, PrivateAttr
+from llama_index.core.bridge.pydantic import BaseModel, Field, PrivateAttr, ConfigDict
 from llama_index.core.callbacks import (
     CallbackManager,
     trace_method,
@@ -55,6 +55,7 @@ class CustomSimpleAgentWorker(BaseModel, BaseAgentWorker):
 
     """
 
+    model_config = ConfigDict(arbitrary_types_allowed=True)
     tools: Sequence[BaseTool] = Field(..., description="Tools to use for reasoning")
     llm: LLM = Field(..., description="LLM to use")
     callback_manager: CallbackManager = Field(
@@ -67,9 +68,6 @@ class CustomSimpleAgentWorker(BaseModel, BaseAgentWorker):
 
     _get_tools: Callable[[str], Sequence[BaseTool]] = PrivateAttr()
 
-    class Config:
-        arbitrary_types_allowed = True
-
     def __init__(
         self,
         tools: Sequence[BaseTool],
@@ -79,18 +77,7 @@ def __init__(
         tool_retriever: Optional[ObjectRetriever[BaseTool]] = None,
         **kwargs: Any,
     ) -> None:
-        if len(tools) > 0 and tool_retriever is not None:
-            raise ValueError("Cannot specify both tools and tool_retriever")
-        elif len(tools) > 0:
-            self._get_tools = lambda _: tools
-        elif tool_retriever is not None:
-            tool_retriever_c = cast(ObjectRetriever[BaseTool], tool_retriever)
-            self._get_tools = lambda message: tool_retriever_c.retrieve(message)
-        else:
-            self._get_tools = lambda _: []
-
         callback_manager = callback_manager or CallbackManager([])
-
         super().__init__(
             tools=tools,
             llm=llm,
@@ -100,6 +87,16 @@ def __init__(
             **kwargs,
         )
 
+        if len(tools) > 0 and tool_retriever is not None:
+            raise ValueError("Cannot specify both tools and tool_retriever")
+        elif len(tools) > 0:
+            self._get_tools = lambda _: tools
+        elif tool_retriever is not None:
+            tool_retriever_c = cast(ObjectRetriever[BaseTool], tool_retriever)
+            self._get_tools = lambda message: tool_retriever_c.retrieve(message)
+        else:
+            self._get_tools = lambda _: []
+
     @classmethod
     def from_tools(
         cls,

llama-index-core/llama_index/core/agent/custom/simple_function.py

@@ -16,7 +16,7 @@
     TaskStep,
     TaskStepOutput,
 )
-from llama_index.core.bridge.pydantic import BaseModel, Field
+from llama_index.core.bridge.pydantic import BaseModel, Field, ConfigDict
 from llama_index.core.callbacks import (
     CallbackManager,
     trace_method,
@@ -44,6 +44,7 @@ class FnAgentWorker(BaseModel, BaseAgentWorker):
 
     """
 
+    model_config = ConfigDict(arbitrary_types_allowed=True)
     fn: Callable = Field(..., description="Function to run.")
     async_fn: Optional[Callable] = Field(
         None, description="Async function to run. If not provided, will run `fn`."
@@ -56,9 +57,6 @@ class FnAgentWorker(BaseModel, BaseAgentWorker):
 
     verbose: bool = Field(False, description="Verbose mode.")
 
-    class Config:
-        arbitrary_types_allowed = True
-
     def __init__(
         self,
         fn: Callable,

llama-index-core/llama_index/core/agent/react/formatter.py

@@ -13,7 +13,7 @@
     ObservationReasoningStep,
 )
 from llama_index.core.base.llms.types import ChatMessage, MessageRole
-from llama_index.core.bridge.pydantic import BaseModel
+from llama_index.core.bridge.pydantic import BaseModel, ConfigDict
 from llama_index.core.tools import BaseTool
 
 logger = logging.getLogger(__name__)
@@ -36,8 +36,7 @@ def get_react_tool_descriptions(tools: Sequence[BaseTool]) -> List[str]:
 class BaseAgentChatFormatter(BaseModel):
     """Base chat formatter."""
 
-    class Config:
-        arbitrary_types_allowed = True
+    model_config = ConfigDict(arbitrary_types_allowed=True)
 
     @abstractmethod
     def format(

llama-index-core/llama_index/core/base/agent/types.py

@@ -7,7 +7,12 @@
 from llama_index.core.base.base_query_engine import BaseQueryEngine
 from llama_index.core.base.llms.types import ChatMessage
 from llama_index.core.base.response.schema import RESPONSE_TYPE, Response
-from llama_index.core.bridge.pydantic import BaseModel, Field
+from llama_index.core.bridge.pydantic import (
+    BaseModel,
+    Field,
+    SerializeAsAny,
+    ConfigDict,
+)
 from llama_index.core.callbacks import CallbackManager, trace_method
 from llama_index.core.chat_engine.types import (
     BaseChatEngine,
@@ -80,11 +85,11 @@ class TaskStep(BaseModel):
 
     """
 
-    task_id: str = Field(..., diescription="Task ID")
+    task_id: str = Field(..., description="Task ID")
     step_id: str = Field(..., description="Step ID")
     input: Optional[str] = Field(default=None, description="User input")
     # memory: BaseMemory = Field(
-    #     ..., type=BaseMemory, description="Conversational Memory"
+    #     ..., description="Conversational Memory"
     # )
     step_state: Dict[str, Any] = Field(
         default_factory=dict, description="Additional state for a given step."
@@ -155,25 +160,22 @@ class Task(BaseModel):
 
     """
 
-    class Config:
-        arbitrary_types_allowed = True
-
+    model_config = ConfigDict(arbitrary_types_allowed=True)
     task_id: str = Field(
-        default_factory=lambda: str(uuid.uuid4()), type=str, description="Task ID"
+        default_factory=lambda: str(uuid.uuid4()), description="Task ID"
     )
-    input: str = Field(..., type=str, description="User input")
+    input: str = Field(..., description="User input")
 
     # NOTE: this is state that may be modified throughout the course of execution of the task
-    memory: BaseMemory = Field(
+    memory: SerializeAsAny[BaseMemory] = Field(
         ...,
-        type=BaseMemory,
         description=(
             "Conversational Memory. Maintains state before execution of this task."
         ),
     )
 
     callback_manager: CallbackManager = Field(
-        default_factory=CallbackManager,
+        default_factory=lambda: CallbackManager([]),
         exclude=True,
         description="Callback manager for the task.",
     )
@@ -190,8 +192,7 @@ class Config:
 class BaseAgentWorker(PromptMixin, DispatcherSpanMixin):
     """Base agent worker."""
 
-    class Config:
-        arbitrary_types_allowed = True
+    model_config = ConfigDict(arbitrary_types_allowed=True)
 
     def _get_prompts(self) -> PromptDictType:
         """Get prompts."""

llama-index-core/llama_index/core/base/base_query_engine.py

@@ -12,7 +12,7 @@
     validate_and_convert_stringable,
 )
 from llama_index.core.base.response.schema import RESPONSE_TYPE
-from llama_index.core.bridge.pydantic import Field
+from llama_index.core.bridge.pydantic import Field, ConfigDict, SerializeAsAny
 from llama_index.core.callbacks.base import CallbackManager
 from llama_index.core.prompts.mixin import PromptDictType, PromptMixin
 from llama_index.core.schema import NodeWithScore, QueryBundle, QueryType
@@ -108,10 +108,10 @@ def _as_query_component(self, **kwargs: Any) -> QueryComponent:
 class QueryEngineComponent(QueryComponent):
     """Query engine component."""
 
-    query_engine: BaseQueryEngine = Field(..., description="Query engine")
-
-    class Config:
-        arbitrary_types_allowed = True
+    model_config = ConfigDict(arbitrary_types_allowed=True)
+    query_engine: SerializeAsAny[BaseQueryEngine] = Field(
+        ..., description="Query engine"
+    )
 
     def set_callback_manager(self, callback_manager: CallbackManager) -> None:
         """Set callback manager."""