-
Notifications
You must be signed in to change notification settings - Fork 19
feat(ChatDeepSeek): Add deepseek support #147
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
Merged
Changes from all commits
Commits
Show all changes
4 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,171 @@ | ||
| from __future__ import annotations | ||
|
|
||
| import os | ||
| from typing import TYPE_CHECKING, Optional, cast | ||
|
|
||
| from ._chat import Chat | ||
| from ._logging import log_model_default | ||
| from ._provider_openai import OpenAIProvider | ||
| from ._turn import Turn | ||
| from ._utils import MISSING, MISSING_TYPE, is_testing | ||
|
|
||
| if TYPE_CHECKING: | ||
| from openai.types.chat import ChatCompletion, ChatCompletionMessageParam | ||
|
|
||
| from .types.openai import ChatClientArgs, SubmitInputArgs | ||
|
|
||
|
|
||
| def ChatDeepSeek( | ||
| *, | ||
| system_prompt: Optional[str] = None, | ||
| model: Optional[str] = None, | ||
| api_key: Optional[str] = None, | ||
| base_url: str = "https://api.deepseek.com", | ||
| seed: Optional[int] | MISSING_TYPE = MISSING, | ||
| kwargs: Optional["ChatClientArgs"] = None, | ||
| ) -> Chat["SubmitInputArgs", ChatCompletion]: | ||
| """ | ||
| Chat with a model hosted on DeepSeek. | ||
|
|
||
| DeepSeek is a platform for AI inference with competitive pricing | ||
| and performance. | ||
|
|
||
| Prerequisites | ||
| ------------- | ||
|
|
||
| ::: {.callout-note} | ||
| ## API key | ||
|
|
||
| Sign up at <https://platform.deepseek.com> to get an API key. | ||
| ::: | ||
|
|
||
| Examples | ||
| -------- | ||
|
|
||
| ```python | ||
| import os | ||
| from chatlas import ChatDeepSeek | ||
|
|
||
| chat = ChatDeepSeek(api_key=os.getenv("DEEPSEEK_API_KEY")) | ||
| chat.chat("What is the capital of France?") | ||
| ``` | ||
|
|
||
| Known limitations | ||
| -------------- | ||
|
|
||
| * Structured data extraction is not supported. | ||
| * Images are not supported. | ||
|
|
||
| Parameters | ||
| ---------- | ||
| system_prompt | ||
| A system prompt to set the behavior of the assistant. | ||
| model | ||
| The model to use for the chat. The default, None, will pick a reasonable | ||
| default, and warn you about it. We strongly recommend explicitly choosing | ||
| a model for all but the most casual use. | ||
| api_key | ||
| The API key to use for authentication. You generally should not supply | ||
| this directly, but instead set the `DEEPSEEK_API_KEY` environment variable. | ||
| base_url | ||
| The base URL to the endpoint; the default uses DeepSeek's API. | ||
| seed | ||
| Optional integer seed that DeepSeek uses to try and make output more | ||
| reproducible. | ||
| kwargs | ||
| Additional arguments to pass to the `openai.OpenAI()` client constructor. | ||
|
|
||
| Returns | ||
| ------- | ||
| Chat | ||
| A chat object that retains the state of the conversation. | ||
|
|
||
| Note | ||
| ---- | ||
| This function is a lightweight wrapper around [](`~chatlas.ChatOpenAI`) with | ||
| the defaults tweaked for DeepSeek. | ||
|
|
||
| Note | ||
| ---- | ||
| Pasting an API key into a chat constructor (e.g., `ChatDeepSeek(api_key="...")`) | ||
| is the simplest way to get started, and is fine for interactive use, but is | ||
| problematic for code that may be shared with others. | ||
|
|
||
| Instead, consider using environment variables or a configuration file to manage | ||
| your credentials. One popular way to manage credentials is to use a `.env` file | ||
| to store your credentials, and then use the `python-dotenv` package to load them | ||
| into your environment. | ||
|
|
||
| ```shell | ||
| pip install python-dotenv | ||
| ``` | ||
|
|
||
| ```shell | ||
| # .env | ||
| DEEPSEEK_API_KEY=... | ||
| ``` | ||
|
|
||
| ```python | ||
| from chatlas import ChatDeepSeek | ||
| from dotenv import load_dotenv | ||
|
|
||
| load_dotenv() | ||
| chat = ChatDeepSeek() | ||
| chat.console() | ||
| ``` | ||
|
|
||
| Another, more general, solution is to load your environment variables into the shell | ||
| before starting Python (maybe in a `.bashrc`, `.zshrc`, etc. file): | ||
|
|
||
| ```shell | ||
| export DEEPSEEK_API_KEY=... | ||
| ``` | ||
| """ | ||
| if model is None: | ||
| model = log_model_default("deepseek-chat") | ||
|
|
||
| if api_key is None: | ||
| api_key = os.getenv("DEEPSEEK_API_KEY") | ||
|
|
||
| if isinstance(seed, MISSING_TYPE): | ||
| seed = 1014 if is_testing() else None | ||
|
|
||
| return Chat( | ||
| provider=DeepSeekProvider( | ||
| api_key=api_key, | ||
| model=model, | ||
| base_url=base_url, | ||
| seed=seed, | ||
| name="DeepSeek", | ||
| kwargs=kwargs, | ||
| ), | ||
| system_prompt=system_prompt, | ||
| ) | ||
|
|
||
|
|
||
| class DeepSeekProvider(OpenAIProvider): | ||
| @staticmethod | ||
| def _as_message_param(turns: list[Turn]) -> list["ChatCompletionMessageParam"]: | ||
| from openai.types.chat import ( | ||
| ChatCompletionAssistantMessageParam, | ||
| ChatCompletionUserMessageParam, | ||
| ) | ||
|
|
||
| params = OpenAIProvider._as_message_param(turns) | ||
|
|
||
| # Content must be a string | ||
| for i, param in enumerate(params): | ||
| if param["role"] in ["assistant", "user"]: | ||
| param = cast( | ||
| ChatCompletionAssistantMessageParam | ||
| | ChatCompletionUserMessageParam, | ||
| param, | ||
| ) | ||
| contents = param.get("content", None) | ||
| if not isinstance(contents, list): | ||
| continue | ||
| params[i]["content"] = "".join( | ||
| content.get("text", "") for content in contents | ||
| ) | ||
|
|
||
| return params | ||
cpsievert marked this conversation as resolved.
Show resolved
Hide resolved
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,62 @@ | ||
| import os | ||
|
|
||
| import pytest | ||
| from chatlas import ChatDeepSeek | ||
|
|
||
| from .conftest import ( | ||
| assert_tools_async, | ||
| assert_tools_simple, | ||
| assert_turns_existing, | ||
| assert_turns_system, | ||
| ) | ||
|
|
||
| api_key = os.getenv("DEEPSEEK_API_KEY") | ||
| if api_key is None: | ||
| pytest.skip("DEEPSEEK_API_KEY is not set; skipping tests", allow_module_level=True) | ||
|
|
||
|
|
||
| def test_deepseek_simple_request(): | ||
| chat = ChatDeepSeek( | ||
| system_prompt="Be as terse as possible; no punctuation", | ||
| ) | ||
| chat.chat("What is 1 + 1?") | ||
| turn = chat.get_last_turn() | ||
| assert turn is not None | ||
| assert turn.tokens is not None | ||
| assert len(turn.tokens) == 3 | ||
| assert turn.tokens[0] >= 10 # More lenient assertion for DeepSeek | ||
| assert turn.finish_reason == "stop" | ||
|
|
||
|
|
||
| @pytest.mark.asyncio | ||
| async def test_deepseek_simple_streaming_request(): | ||
| chat = ChatDeepSeek( | ||
| system_prompt="Be as terse as possible; no punctuation", | ||
| ) | ||
| res = [] | ||
| async for x in await chat.stream_async("What is 1 + 1?"): | ||
| res.append(x) | ||
| assert "2" in "".join(res) | ||
| turn = chat.get_last_turn() | ||
| assert turn is not None | ||
| assert turn.finish_reason == "stop" | ||
|
|
||
|
|
||
| def test_deepseek_respects_turns_interface(): | ||
| chat_fun = ChatDeepSeek | ||
| assert_turns_system(chat_fun) | ||
| assert_turns_existing(chat_fun) | ||
|
|
||
|
|
||
| def test_deepseek_tool_variations(): | ||
| chat_fun = ChatDeepSeek | ||
| assert_tools_simple(chat_fun) | ||
|
|
||
|
|
||
| @pytest.mark.asyncio | ||
| async def test_deepseek_tool_variations_async(): | ||
| chat_fun = ChatDeepSeek | ||
| await assert_tools_async(chat_fun) | ||
|
|
||
|
|
||
| # Doesn't seem to support data extraction or images |
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.