Conversation
在.env.example中添加了XINFERENCE_OPENAI_ENDPOINT和XINFERENCE_MODEL环境变量,并在src/utils/utils.py中实现了对Xinference模型的支持。新增的模型名称列表包括qwen2.5和deepseek系列模型。
|
lao seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account. You have signed the CLA already but the status is still pending? Let us recheck it. |
There was a problem hiding this comment.
mrge found 4 issues across 2 files. View them in mrge.io
| SiliconFLOW_API_KEY= | ||
|
|
||
| XINFERENCE_OPENAI_ENDPOINT=https://api.xinference.com/v1 | ||
| XINFERENCE_MODEL= |
There was a problem hiding this comment.
Missing documentation for expected values for XINFERENCE_MODEL environment variable
| model_name=kwargs.get("model_name", "Qwen/QwQ-32B"), | ||
| temperature=kwargs.get("temperature", 0.0), | ||
| ) | ||
| elif provider == "xinference": |
There was a problem hiding this comment.
Missing entry for "xinference" in PROVIDER_DISPLAY_NAMES dictionary
src/utils/utils.py
Outdated
| base_url = kwargs.get("base_url") | ||
|
|
||
| return ChatOpenAI( | ||
| model=kwargs.get("model_name", "gpt-4o"), |
There was a problem hiding this comment.
Default model for Xinference provider is set to "gpt-4o" which is not in the xinference model list
| temperature=kwargs.get("temperature", 0.0), | ||
| ) | ||
| elif provider == "xinference": | ||
| if not kwargs.get("base_url", ""): |
There was a problem hiding this comment.
API key retrieval uses a different environment variable pattern than what's defined in .env.example
在utils.py中添加对XInference的支持,并在.env.example中添加XINFERENCE_API_KEY环境变量。这些更改使得系统能够更好地集成XInference服务。
Summary by mrge
Added support for Xinference models including Qwen2.5 and Deepseek series. This integration enables users to leverage these models through a standardized interface with configurable endpoints.
New Features
Refactors
在.env.example中添加了XINFERENCE_OPENAI_ENDPOINT和XINFERENCE_MODEL环境变量,并在src/utils/utils.py中实现了对Xinference模型的支持。新增的模型名称列表包括qwen2.5和deepseek系列模型。