Skip to content

armgit / camel

Public Imported from GitHub
0

🐫 CAMEL: The first and the best multi-agent framework. Finding the Scaling Law of Agents. https://www.camel-ai.org

8.1 KB
Raw Blame
<a id="camel.models.openai_compatible_model"></a>

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel"></a>

## OpenAICompatibleModel

```python
class OpenAICompatibleModel(BaseModelBackend):
```

Constructor for model backend supporting OpenAI compatibility.

**Parameters:**

- **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:obj:`openai.ChatCompletion.create()`. If :obj:`None`, :obj:`{}` will be used. (default: :obj:`None`)
- **api_key** (str): The API key for authenticating with the model service.
- **url** (str): The url to the model service.
- **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`)
- **timeout** (Optional[float], optional): The timeout value in seconds for API calls. If not provided, will fall back to the MODEL_TIMEOUT environment variable or default to 180 seconds. (default: :obj:`None`)
- **max_retries** (int, optional): Maximum number of retries for API calls. (default: :obj:`3`)
- **client** (Optional[Any], optional): A custom synchronous OpenAI-compatible client instance. If provided, this client will be used instead of creating a new one. Useful for RL frameworks like AReaL or rLLM that provide OpenAI-compatible clients (e.g., ArealOpenAI). The client should implement the OpenAI client interface with `.chat.completions.create()` and `.beta.chat. completions.parse()` methods. (default: :obj:`None`)
- **async_client** (Optional[Any], optional): A custom asynchronous OpenAI-compatible client instance. If provided, this client will be used instead of creating a new one. The client should implement the AsyncOpenAI client interface. (default: :obj:`None`)
- **api_mode** (`Literal["chat_completions", "responses"], optional`): API mode to use. Supported values: `"chat_completions"` (default) and `"responses"`. **kwargs (Any): Additional arguments to pass to the OpenAI client initialization. These can include parameters like 'organization', 'default_headers', 'http_client', etc. Ignored if custom clients are provided.

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel.__init__"></a>

### __init__

```python
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,
    timeout: Optional[float] = None,
    max_retries: int = 3,
    client: Optional[Any] = None,
    async_client: Optional[Any] = None,
    api_mode: Literal['chat_completions', 'responses'] = 'chat_completions',
    **kwargs: Any
):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._run"></a>

### _run

```python
def _run(
    self,
    messages: List[OpenAIMessage],
    response_format: Optional[Type[BaseModel]] = None,
    tools: Optional[List[Dict[str, Any]]] = None
):
```

Runs inference of OpenAI chat completion.

**Parameters:**

- **messages** (List[OpenAIMessage]): Message list with the chat history in OpenAI API format.
- **response_format** (Optional[Type[BaseModel]]): The format of the response.
- **tools** (Optional[List[Dict[str, Any]]]): The schema of the tools to use for the request.

**Returns:**

  Union[ChatCompletion, Stream[ChatCompletionChunk]]:
`ChatCompletion` in the non-stream mode, or
`Stream[ChatCompletionChunk]` in the stream mode.
`ChatCompletionStreamManager[BaseModel]` for
structured output streaming.

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._request_chat_completion"></a>

### _request_chat_completion

```python
def _request_chat_completion(
    self,
    messages: List[OpenAIMessage],
    tools: Optional[List[Dict[str, Any]]] = None
):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._request_parse"></a>

### _request_parse

```python
def _request_parse(
    self,
    messages: List[OpenAIMessage],
    response_format: Type[BaseModel],
    tools: Optional[List[Dict[str, Any]]] = None
):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._request_stream_parse"></a>

### _request_stream_parse

```python
def _request_stream_parse(
    self,
    messages: List[OpenAIMessage],
    response_format: Type[BaseModel],
    tools: Optional[List[Dict[str, Any]]] = None
):
```

Request streaming structured output parsing.

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._normalize_tools_for_responses_api"></a>

### _normalize_tools_for_responses_api

```python
def _normalize_tools_for_responses_api(tools: List[Dict[str, Any]]):
```

Convert chat-completions style function tools to Responses style.

Input:
`{"type":"function","function":{"name":"foo",...}}`
Output:
`{"type":"function","name":"foo",...}`

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._enforce_object_additional_properties_false"></a>

### _enforce_object_additional_properties_false

```python
def _enforce_object_additional_properties_false(schema: Any):
```

Recursively enforce strict object schema for Responses API.

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._get_response_chain_session_key"></a>

### _get_response_chain_session_key

```python
def _get_response_chain_session_key(self):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._clear_response_chain_state"></a>

### _clear_response_chain_state

```python
def _clear_response_chain_state(self, session_key: str):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._responses_chain_enabled"></a>

### _responses_chain_enabled

```python
def _responses_chain_enabled(request_config: Dict[str, Any]):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._prepare_responses_input_and_chain"></a>

### _prepare_responses_input_and_chain

```python
def _prepare_responses_input_and_chain(self, messages: List[OpenAIMessage], chain_enabled: bool = True):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._convert_messages_to_responses_input"></a>

### _convert_messages_to_responses_input

```python
def _convert_messages_to_responses_input(messages: List[OpenAIMessage]):
```

Convert chat-completions style messages to Responses input items.

This specifically rewrites tool-calling history:
- assistant message with `tool_calls` -> one or more `function_call`
items (+ optional assistant message when content exists)
- tool message -> `function_call_output` item

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._save_response_chain_state"></a>

### _save_response_chain_state

```python
def _save_response_chain_state(
    self,
    session_key: str,
    response_id: Optional[str],
    message_count: int
):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._prepare_responses_request_config"></a>

### _prepare_responses_request_config

```python
def _prepare_responses_request_config(
    self,
    tools: Optional[List[Dict[str, Any]]] = None,
    response_format: Optional[Type[BaseModel]] = None,
    stream: bool = False
):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._request_responses"></a>

### _request_responses

```python
def _request_responses(
    self,
    messages: List[OpenAIMessage],
    response_format: Optional[Type[BaseModel]],
    tools: Optional[List[Dict[str, Any]]] = None
):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel._request_responses_stream"></a>

### _request_responses_stream

```python
def _request_responses_stream(
    self,
    messages: List[OpenAIMessage],
    response_format: Optional[Type[BaseModel]],
    tools: Optional[List[Dict[str, Any]]] = None
):
```

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel.token_counter"></a>

### token_counter

```python
def token_counter(self):
```

**Returns:**

  OpenAITokenCounter: The token counter following the model's
tokenization style.

<a id="camel.models.openai_compatible_model.OpenAICompatibleModel.stream"></a>

### stream

```python
def stream(self):
```

**Returns:**

  bool: Whether the model is in stream mode.