Merge pull request #95 from mistralai/release/v0.3.0

release 0.3.0: add support for completion

Author: jean-malo

examples/async_completion.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python
+
+import asyncio
+import os
+
+from mistralai.async_client import MistralAsyncClient
+
+
+async def main():
+    api_key = os.environ["MISTRAL_API_KEY"]
+
+    client = MistralAsyncClient(api_key=api_key)
+
+    prompt = "def fibonacci(n: int):"
+    suffix = "n = int(input('Enter a number: '))\nprint(fibonacci(n))"
+
+    response = await client.completion(
+        model="codestral-latest",
+        prompt=prompt,
+        suffix=suffix,
+    )
+
+    print(
+        f"""
+{prompt}
+{response.choices[0].message.content}
+{suffix}
+"""
+    )
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/chatbot_with_streaming.py

@@ -12,11 +12,12 @@
 from mistralai.models.chat_completion import ChatMessage
 
 MODEL_LIST = [
-    "mistral-tiny",
-    "mistral-small",
-    "mistral-medium",
+    "mistral-small-latest",
+    "mistral-medium-latest",
+    "mistral-large-latest",
+    "codestral-latest",
 ]
-DEFAULT_MODEL = "mistral-small"
+DEFAULT_MODEL = "mistral-small-latest"
 DEFAULT_TEMPERATURE = 0.7
 LOG_FORMAT = "%(asctime)s - %(levelname)s - %(message)s"
 # A dictionary of all commands and their arguments, used for tab completion.

examples/code_completion.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python
+
+import asyncio
+import os
+
+from mistralai.client import MistralClient
+
+
+async def main():
+    api_key = os.environ["MISTRAL_API_KEY"]
+
+    client = MistralClient(api_key=api_key)
+
+    prompt = "def fibonacci(n: int):"
+    suffix = "n = int(input('Enter a number: '))\nprint(fibonacci(n))"
+
+    response = client.completion(
+        model="codestral-latest",
+        prompt=prompt,
+        suffix=suffix,
+    )
+
+    print(
+        f"""
+{prompt}
+{response.choices[0].message.content}
+{suffix}
+"""
+    )
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/completion_with_streaming.py

@@ -0,0 +1,29 @@
+#!/usr/bin/env python
+
+import asyncio
+import os
+
+from mistralai.client import MistralClient
+
+
+async def main():
+    api_key = os.environ["MISTRAL_API_KEY"]
+
+    client = MistralClient(api_key=api_key)
+
+    prompt = "def fibonacci(n: int):"
+    suffix = "n = int(input('Enter a number: '))\nprint(fibonacci(n))"
+
+    print(prompt)
+    for chunk in client.completion_stream(
+        model="codestral-latest",
+        prompt=prompt,
+        suffix=suffix,
+    ):
+        if chunk.choices[0].delta.content is not None:
+            print(chunk.choices[0].delta.content, end="")
+    print(suffix)
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "mistralai"
-version = "0.2.0"
+version = "0.3.0"
 description = ""
 authors = ["Bam4d <[email protected]>"]
 readme = "README.md"

src/mistralai/async_client.py

@@ -92,7 +92,7 @@ async def _check_response(self, response: Response) -> Dict[str, Any]:
     async def _request(
         self,
         method: str,
-        json: Dict[str, Any],
+        json: Optional[Dict[str, Any]],
         path: str,
         stream: bool = False,
         attempt: int = 1,
@@ -291,3 +291,74 @@ async def list_models(self) -> ModelList:
             return ModelList(**response)
 
         raise MistralException("No response received")
+
+    async def completion(
+        self,
+        model: str,
+        prompt: str,
+        suffix: Optional[str] = None,
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
+        top_p: Optional[float] = None,
+        random_seed: Optional[int] = None,
+        stop: Optional[List[str]] = None,
+    ) -> ChatCompletionResponse:
+        """An asynchronous completion endpoint that returns a single response.
+
+        Args:
+            model (str): model the name of the model to get completions with, e.g. codestral-latest
+            prompt (str): the prompt to complete
+            suffix (Optional[str]): the suffix to append to the prompt for fill-in-the-middle completion
+            temperature (Optional[float], optional): temperature the temperature to use for sampling, e.g. 0.5.
+            max_tokens (Optional[int], optional): the maximum number of tokens to generate, e.g. 100. Defaults to None.
+            top_p (Optional[float], optional): the cumulative probability of tokens to generate, e.g. 0.9.
+            Defaults to None.
+            random_seed (Optional[int], optional): the random seed to use for sampling, e.g. 42. Defaults to None.
+            stop (Optional[List[str]], optional): a list of tokens to stop generation at, e.g. ['/n/n']
+        Returns:
+            Dict[str, Any]: a response object containing the generated text.
+        """
+        request = self._make_completion_request(
+            prompt, model, suffix, temperature, max_tokens, top_p, random_seed, stop
+        )
+        single_response = self._request("post", request, "v1/fim/completions")
+
+        async for response in single_response:
+            return ChatCompletionResponse(**response)
+
+        raise MistralException("No response received")
+
+    async def completion_stream(
+        self,
+        model: str,
+        prompt: str,
+        suffix: Optional[str] = None,
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
+        top_p: Optional[float] = None,
+        random_seed: Optional[int] = None,
+        stop: Optional[List[str]] = None,
+    ) -> AsyncGenerator[ChatCompletionStreamResponse, None]:
+        """An asynchronous completion endpoint that returns a streaming response.
+
+        Args:
+            model (str): model the name of the model to get completions with, e.g. codestral-latest
+            prompt (str): the prompt to complete
+            suffix (Optional[str]): the suffix to append to the prompt for fill-in-the-middle completion
+            temperature (Optional[float], optional): temperature the temperature to use for sampling, e.g. 0.5.
+            max_tokens (Optional[int], optional): the maximum number of tokens to generate, e.g. 100. Defaults to None.
+            top_p (Optional[float], optional): the cumulative probability of tokens to generate, e.g. 0.9.
+            Defaults to None.
+            random_seed (Optional[int], optional): the random seed to use for sampling, e.g. 42. Defaults to None.
+            stop (Optional[List[str]], optional): a list of tokens to stop generation at, e.g. ['/n/n']
+
+        Returns:
+            Dict[str, Any]: a response object containing the generated text.
+        """
+        request = self._make_completion_request(
+            prompt, model, suffix, temperature, max_tokens, top_p, random_seed, stop, stream=True
+        )
+        async_response = self._request("post", request, "v1/fim/completions", stream=True)
+
+        async for json_response in async_response:
+            yield ChatCompletionStreamResponse(**json_response)

src/mistralai/client.py

@@ -85,7 +85,7 @@ def _check_response(self, response: Response) -> Dict[str, Any]:
     def _request(
         self,
         method: str,
-        json: Dict[str, Any],
+        json: Optional[Dict[str, Any]],
         path: str,
         stream: bool = False,
         attempt: int = 1,
@@ -285,3 +285,77 @@ def list_models(self) -> ModelList:
             return ModelList(**response)
 
         raise MistralException("No response received")
+
+    def completion(
+        self,
+        model: str,
+        prompt: str,
+        suffix: Optional[str] = None,
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
+        top_p: Optional[float] = None,
+        random_seed: Optional[int] = None,
+        stop: Optional[List[str]] = None,
+    ) -> ChatCompletionResponse:
+        """A completion endpoint that returns a single response.
+
+        Args:
+            model (str): model the name of the model to get completion with, e.g. codestral-latest
+            prompt (str): the prompt to complete
+            suffix (Optional[str]): the suffix to append to the prompt for fill-in-the-middle completion
+            temperature (Optional[float], optional): temperature the temperature to use for sampling, e.g. 0.5.
+            max_tokens (Optional[int], optional): the maximum number of tokens to generate, e.g. 100. Defaults to None.
+            top_p (Optional[float], optional): the cumulative probability of tokens to generate, e.g. 0.9.
+            Defaults to None.
+            random_seed (Optional[int], optional): the random seed to use for sampling, e.g. 42. Defaults to None.
+            stop (Optional[List[str]], optional): a list of tokens to stop generation at, e.g. ['/n/n']
+
+        Returns:
+            Dict[str, Any]: a response object containing the generated text.
+        """
+        request = self._make_completion_request(
+            prompt, model, suffix, temperature, max_tokens, top_p, random_seed, stop
+        )
+
+        single_response = self._request("post", request, "v1/fim/completions", stream=False)
+
+        for response in single_response:
+            return ChatCompletionResponse(**response)
+
+        raise MistralException("No response received")
+
+    def completion_stream(
+        self,
+        model: str,
+        prompt: str,
+        suffix: Optional[str] = None,
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
+        top_p: Optional[float] = None,
+        random_seed: Optional[int] = None,
+        stop: Optional[List[str]] = None,
+    ) -> Iterable[ChatCompletionStreamResponse]:
+        """An asynchronous completion endpoint that streams responses.
+
+        Args:
+            model (str): model the name of the model to get completions with, e.g. codestral-latest
+            prompt (str): the prompt to complete
+            suffix (Optional[str]): the suffix to append to the prompt for fill-in-the-middle completion
+            temperature (Optional[float], optional): temperature the temperature to use for sampling, e.g. 0.5.
+            max_tokens (Optional[int], optional): the maximum number of tokens to generate, e.g. 100. Defaults to None.
+            top_p (Optional[float], optional): the cumulative probability of tokens to generate, e.g. 0.9.
+            Defaults to None.
+            random_seed (Optional[int], optional): the random seed to use for sampling, e.g. 42. Defaults to None.
+            stop (Optional[List[str]], optional): a list of tokens to stop generation at, e.g. ['/n/n']
+
+        Returns:
+            Iterable[Dict[str, Any]]: a generator that yields response objects containing the generated text.
+        """
+        request = self._make_completion_request(
+            prompt, model, suffix, temperature, max_tokens, top_p, random_seed, stop, stream=True
+        )
+
+        response = self._request("post", request, "v1/fim/completions", stream=True)
+
+        for json_streamed_response in response:
+            yield ChatCompletionStreamResponse(**json_streamed_response)

src/mistralai/client_base.py

@@ -73,6 +73,63 @@ def _parse_messages(self, messages: List[Any]) -> List[Dict[str, Any]]:
 
         return parsed_messages
 
+    def _make_completion_request(
+        self,
+        prompt: str,
+        model: Optional[str] = None,
+        suffix: Optional[str] = None,
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
+        top_p: Optional[float] = None,
+        random_seed: Optional[int] = None,
+        stop: Optional[List[str]] = None,
+        stream: Optional[bool] = False,
+    ) -> Dict[str, Any]:
+        request_data: Dict[str, Any] = {
+            "prompt": prompt,
+            "suffix": suffix,
+            "model": model,
+            "stream": stream,
+        }
+
+        if stop is not None:
+            request_data["stop"] = stop
+
+        if model is not None:
+            request_data["model"] = model
+        else:
+            if self._default_model is None:
+                raise MistralException(message="model must be provided")
+            request_data["model"] = self._default_model
+
+        request_data.update(
+            self._build_sampling_params(
+                temperature=temperature, max_tokens=max_tokens, top_p=top_p, random_seed=random_seed
+            )
+        )
+
+        self._logger.debug(f"Completion request: {request_data}")
+
+        return request_data
+
+    def _build_sampling_params(
+        self,
+        max_tokens: Optional[int],
+        random_seed: Optional[int],
+        temperature: Optional[float],
+        top_p: Optional[float],
+    ) -> Dict[str, Any]:
+        params = {}
+        if temperature is not None:
+            params["temperature"] = temperature
+        if max_tokens is not None:
+            params["max_tokens"] = max_tokens
+        if top_p is not None:
+            params["top_p"] = top_p
+        if random_seed is not None:
+            params["random_seed"] = random_seed
+        return params
+
     def _make_chat_request(
         self,
         messages: List[Any],
@@ -99,16 +156,14 @@ def _make_chat_request(
                 raise MistralException(message="model must be provided")
             request_data["model"] = self._default_model
 
+        request_data.update(
+            self._build_sampling_params(
+                temperature=temperature, max_tokens=max_tokens, top_p=top_p, random_seed=random_seed
+            )
+        )
+
         if tools is not None:
             request_data["tools"] = self._parse_tools(tools)
-        if temperature is not None:
-            request_data["temperature"] = temperature
-        if max_tokens is not None:
-            request_data["max_tokens"] = max_tokens
-        if top_p is not None:
-            request_data["top_p"] = top_p
-        if random_seed is not None:
-            request_data["random_seed"] = random_seed
         if stream is not None:
             request_data["stream"] = stream