feat: Integrate siliconflow model (#1541)

Co-authored-by: Wendong-Fan <[email protected]>
Co-authored-by: Wendong <[email protected]>

.github/workflows/build_package.yml

@@ -80,6 +80,7 @@ jobs:
           DISCORD_BOT_TOKEN: "${{ secrets.DISCORD_BOT_TOKEN }}"
           INTERNLM_API_KEY: "${{ secrets.INTERNLM_API_KEY }}"
           JINA_API_KEY: "${{ secrets.JINA_API_KEY }}"
+          SILICONFLOW_API_KEY: "${{ secrets.SILICONFLOW_API_KEY }}"
           MOONSHOT_API_KEY: "${{ secrets.MOONSHOT_API_KEY }}"
         run: |
           source venv/bin/activate

.github/workflows/pytest_apps.yml

@@ -31,6 +31,7 @@ jobs:
         COHERE_API_KEY: "${{ secrets.COHERE_API_KEY }}"
         INTERNLM_API_KEY: "${{ secrets.INTERNLM_API_KEY }}"
         MOONSHOT_API_KEY: "${{ secrets.MOONSHOT_API_KEY }}"
+        SILICONFLOW_API_KEY: "${{ secrets.SILICONFLOW_API_KEY }}"
       run: poetry run pytest -v apps/
 
   pytest_examples:
@@ -51,4 +52,5 @@ jobs:
         COHERE_API_KEY: "${{ secrets.COHERE_API_KEY }}"
         INTERNLM_API_KEY: "${{ secrets.INTERNLM_API_KEY }}"
         MOONSHOT_API_KEY: "${{ secrets.MOONSHOT_API_KEY }}"
+        SILICONFLOW_API_KEY: "${{ secrets.SILICONFLOW_API_KEY }}"
       run: poetry run pytest -v examples/

.github/workflows/pytest_package.yml

@@ -60,6 +60,7 @@ jobs:
         INTERNLM_API_KEY: "${{ secrets.INTERNLM_API_KEY }}"
         JINA_API_KEY: "${{ secrets.JINA_API_KEY }}"
         MOONSHOT_API_KEY: "${{ secrets.MOONSHOT_API_KEY }}"
+        SILICONFLOW_API_KEY: "${{ secrets.SILICONFLOW_API_KEY }}"
       run: poetry run pytest --fast-test-mode test/
 
   pytest_package_llm_test:
@@ -109,6 +110,7 @@ jobs:
         INTERNLM_API_KEY: "${{ secrets.INTERNLM_API_KEY }}"
         JINA_API_KEY: "${{ secrets.JINA_API_KEY }}"
         MOONSHOT_API_KEY: "${{ secrets.MOONSHOT_API_KEY }}"
+        SILICONFLOW_API_KEY: "${{ secrets.SILICONFLOW_API_KEY }}"
       run: poetry run pytest --llm-test-only test/
 
   pytest_package_very_slow_test:

camel/configs/__init__.py

@@ -33,6 +33,7 @@
     SambaVerseAPIConfig,
 )
 from .sglang_config import SGLANG_API_PARAMS, SGLangConfig
+from .siliconflow_config import SILICONFLOW_API_PARAMS, SiliconFlowConfig
 from .togetherai_config import TOGETHERAI_API_PARAMS, TogetherAIConfig
 from .vllm_config import VLLM_API_PARAMS, VLLMConfig
 from .yi_config import YI_API_PARAMS, YiConfig
@@ -82,4 +83,6 @@
     'INTERNLM_API_PARAMS',
     'MoonshotConfig',
     "MOONSHOT_API_PARAMS",
+    'SiliconFlowConfig',
+    'SILICONFLOW_API_PARAMS',
 ]

camel/configs/siliconflow_config.py

@@ -0,0 +1,91 @@
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+from __future__ import annotations
+
+from typing import Any, Sequence, Type, Union
+
+from pydantic import BaseModel
+
+from camel.configs.base_config import BaseConfig
+from camel.types import NOT_GIVEN, NotGiven
+
+
+class SiliconFlowConfig(BaseConfig):
+    r"""Defines the parameters for generating chat completions using the
+    SiliconFlow API.
+
+    Args:
+        temperature (float, optional): Determines the degree of randomness
+            in the response. (default: :obj:`0.7`)
+        top_p (float, optional): The top_p (nucleus) parameter is used to
+            dynamically adjust the number of choices for each predicted token
+            based on the cumulative probabilities. (default: :obj:`0.7`)
+        n (int, optional): Number of generations to return. (default::obj:`1`)
+        response_format (object, optional): An object specifying the format
+            that the model must output.
+        stream (bool, optional): If set, tokens are returned as Server-Sent
+            Events as they are made available. (default: :obj:`False`)
+        stop (str or list, optional): Up to :obj:`4` sequences where the API
+            will stop generating further tokens. (default: :obj:`None`)
+        max_tokens (int, optional): The maximum number of tokens to generate.
+            (default: :obj:`None`)
+        frequency_penalty (float, optional): Number between :obj:`-2.0` and
+            :obj:`2.0`. Positive values penalize new tokens based on their
+            existing frequency in the text so far, decreasing the model's
+            likelihood to repeat the same line verbatim. See more information
+            about frequency and presence penalties. (default: :obj:`0.0`)
+        tools (list[FunctionTool], optional): A list of tools the model may
+            call. Currently, only functions are supported as a tool. Use this
+            to provide a list of functions the model may generate JSON inputs
+            for. A max of 128 functions are supported.
+    """
+
+    temperature: float = 0.7
+    top_p: float = 0.7
+    n: int = 1
+    stream: bool = False
+    stop: Union[str, Sequence[str], NotGiven] = NOT_GIVEN
+    max_tokens: Union[int, NotGiven] = NOT_GIVEN
+    response_format: Union[Type[BaseModel], dict, NotGiven] = NOT_GIVEN
+    frequency_penalty: float = 0.0
+
+    def as_dict(self) -> dict[str, Any]:
+        r"""Convert the current configuration to a dictionary.
+
+        This method converts the current configuration object to a dictionary
+        representation, which can be used for serialization or other purposes.
+
+        Returns:
+            dict[str, Any]: A dictionary representation of the current
+                configuration.
+        """
+        config_dict = self.model_dump()
+        if self.tools:
+            from camel.toolkits import FunctionTool
+
+            tools_schema = []
+            for tool in self.tools:
+                if not isinstance(tool, FunctionTool):
+                    raise ValueError(
+                        f"The tool {tool} should "
+                        "be an instance of `FunctionTool`."
+                    )
+                tools_schema.append(tool.get_openai_tool_schema())
+        config_dict["tools"] = NOT_GIVEN
+        return config_dict
+
+
+SILICONFLOW_API_PARAMS = {
+    param for param in SiliconFlowConfig.model_fields.keys()
+}

camel/embeddings/jina_embedding.py

@@ -35,6 +35,11 @@ class JinaEmbedding(BaseEmbedding[Union[str, Image.Image]]):
             Jina AI. (default: :obj:`None`)
         dimensions (Optional[int], optional): The dimension of the output
             embeddings. (default: :obj:`None`)
+        embedding_type (Optional[str], optional): The type of embedding format
+            to generate. Options: 'int8' (binary encoding with higher storage
+            and transfer efficiency), 'uint8' (unsigned binary encoding with
+            higher storage and transfer efficiency), 'base64' (base64 string
+            encoding with higher transfer efficiency). (default: :obj:`None`)
         task (Optional[str], optional): The type of task for text embeddings.
             Options: retrieval.query, retrieval.passage, text-matching,
             classification, separation. (default: :obj:`None`)
@@ -120,7 +125,7 @@ def embed_list(
             else:
                 raise ValueError(
                     f"Input type {type(obj)} is not supported. "
-                    "Must be either str or PIL.Image"
+                    "Must be either str or PIL.Image."
                 )
 
         data = {

camel/models/model_factory.py

@@ -32,6 +32,7 @@
 from camel.models.reka_model import RekaModel
 from camel.models.samba_model import SambaModel
 from camel.models.sglang_model import SGLangModel
+from camel.models.siliconflow_model import SiliconFlowModel
 from camel.models.stub_model import StubModel
 from camel.models.togetherai_model import TogetherAIModel
 from camel.models.vllm_model import VLLMModel
@@ -101,6 +102,8 @@ def create(
             model_class = LiteLLMModel
         elif model_platform.is_nvidia:
             model_class = NvidiaModel
+        elif model_platform.is_siliconflow:
+            model_class = SiliconFlowModel
 
         elif model_platform.is_openai and model_type.is_openai:
             model_class = OpenAIModel

camel/models/siliconflow_model.py

@@ -0,0 +1,142 @@
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+import os
+from typing import Any, Dict, List, Optional, Union
+
+from openai import OpenAI, Stream
+
+from camel.configs import SILICONFLOW_API_PARAMS, SiliconFlowConfig
+from camel.messages import OpenAIMessage
+from camel.models import BaseModelBackend
+from camel.types import (
+    ChatCompletion,
+    ChatCompletionChunk,
+    ModelType,
+)
+from camel.utils import (
+    BaseTokenCounter,
+    OpenAITokenCounter,
+    api_keys_required,
+)
+
+
+class SiliconFlowModel(BaseModelBackend):
+    r"""SiliconFlow API in a unified BaseModelBackend interface.
+
+    Args:
+        model_type (Union[ModelType, str]): Model for which a backend is
+            created.
+        model_config_dict (Optional[Dict[str, Any]], optional): A dictionary
+            that will be fed into OpenAI client. If :obj:`None`,
+            :obj:`SiliconFlowConfig().as_dict()` will be used.
+            (default: :obj:`None`)
+        api_key (Optional[str], optional): The API key for authenticating with
+            the SiliconFlow service. (default: :obj:`None`)
+        url (Optional[str], optional): The URL to the SiliconFlow service. If
+            not provided, :obj:`https://api.siliconflow.cn/v1/` will be used.
+            (default: :obj:`None`)
+        token_counter (Optional[BaseTokenCounter], optional): Token counter to
+            use for the model. If not provided, :obj:`OpenAITokenCounter(
+            ModelType.GPT_4O_MINI)` will be used.
+            (default: :obj:`None`)
+    """
+
+    @api_keys_required(
+        [
+            ("api_key", 'SILICONFLOW_API_KEY'),
+        ]
+    )
+    def __init__(
+        self,
+        model_type: Union[ModelType, str],
+        model_config_dict: Optional[Dict[str, Any]] = None,
+        api_key: Optional[str] = None,
+        url: Optional[str] = None,
+        token_counter: Optional[BaseTokenCounter] = None,
+    ) -> None:
+        if model_config_dict is None:
+            model_config_dict = SiliconFlowConfig().as_dict()
+        api_key = api_key or os.environ.get("SILICONFLOW_API_KEY")
+        url = url or os.environ.get(
+            "SILICONFLOW_API_BASE_URL",
+            "https://api.siliconflow.cn/v1/",
+        )
+        super().__init__(
+            model_type, model_config_dict, api_key, url, token_counter
+        )
+        self._client = OpenAI(
+            timeout=180,
+            max_retries=3,
+            api_key=self._api_key,
+            base_url=self._url,
+        )
+
+    def run(
+        self,
+        messages: List[OpenAIMessage],
+    ) -> Union[ChatCompletion, Stream[ChatCompletionChunk]]:
+        r"""Runs inference of SiliconFlow chat completion.
+
+        Args:
+            messages (List[OpenAIMessage]): Message list with the chat history
+                in OpenAI API format.
+
+        Returns:
+            Union[ChatCompletion, Stream[ChatCompletionChunk]]:
+                `ChatCompletion` in the non-stream mode, or
+                `Stream[ChatCompletionChunk]` in the stream mode.
+        """
+        response = self._client.chat.completions.create(
+            messages=messages,
+            model=self.model_type,
+            **self.model_config_dict,
+        )
+        return response
+
+    @property
+    def token_counter(self) -> BaseTokenCounter:
+        r"""Initialize the token counter for the model backend.
+
+        Returns:
+            BaseTokenCounter: The token counter following the model's
+                tokenization style.
+        """
+        if not self._token_counter:
+            self._token_counter = OpenAITokenCounter(ModelType.GPT_4O_MINI)
+        return self._token_counter
+
+    def check_model_config(self):
+        r"""Check whether the model configuration contains any
+        unexpected arguments to SiliconFlow API.
+
+        Raises:
+            ValueError: If the model configuration dictionary contains any
+                unexpected arguments to SiliconFlow API.
+        """
+        for param in self.model_config_dict:
+            if param not in SILICONFLOW_API_PARAMS:
+                raise ValueError(
+                    f"Unexpected argument `{param}` is "
+                    "input into SiliconFlow model backend."
+                )
+
+    @property
+    def stream(self) -> bool:
+        """Returns whether the model is in stream mode, which sends partial
+        results each time.
+
+        Returns:
+            bool: Whether the model is in stream mode.
+        """
+        return self.model_config_dict.get('stream', False)