tingjun-cs

Results 8 comments of tingjun-cs

> DeepSeek R1 does not support tool calling. But QwQ-32B supports it. Ref [#14490](https://github.com/vllm-project/vllm/issues/14490) What about DeepSeek-V3? From the DeepSeek official website, it appears that DeepSeek-Chat (also known as DeepSeek-V3)...

> Yes, v3 supports it. > > /cc [@WangErXiao](https://github.com/WangErXiao) So, what tool-call-parser should be used for DeepSeek-V3?

> I do not think we support it now. Is there a roadmap or schedule for supporting the tool call feature for DeepSeek-V3? If the official team supports this feature,...

> [@tingjun-cs](https://github.com/tingjun-cs) And, welcome contributions if you are interested in it 😄 Can you briefly introduce the implementation principles and ideas in this area? I'm currently just starting to get...

> Yes, it also need to handle logic of serving_chat.py. Currently tools info are in request.tools field. But DeepSeek V3 chat_template only handle tools info of message. [@aarnphm](https://github.com/aarnphm) @WangErXiao @aarnphm...

> Yes, it also need to handle logic of serving_chat.py. Currently tools info are in request.tools field. But DeepSeek V3 chat_template only handle tools info of message. [@aarnphm](https://github.com/aarnphm) @WangErXiao @aarnphm...

> both v3 and r1 uses the same chat template format for tools. from users perspective it should be the same as OpenAI interfaces. > > from vLLM perspective, we...

@uzhilinsky @kpertsch Hi, could you please take a look at this issue when you have time? It's directly blocking our fine-tuning of the Pi0.5 model.