Yuge Zhang comments

Results 279 comments of


                                            Yuge Zhang

Semantic Convention

/ci

Empty triplets when using `with_structured_output` in a langchain agent

I think the problem is related to vllm. If you are using vllm >= 0.10.2, please verify whether return_token_ids + structured output (called guided decoding) works vllm's native chat completion...

Empty triplets when using `with_structured_output` in a langchain agent

It's possible that structured output didn't use the chat completion, thus won't be captured at client side. But since the proxy server side is universal, the raw_gen_ai_request always exist. In...

Empty triplets when using `with_structured_output` in a langchain agent

Yes. You can grep the example folder for examples.

trnasition-level reward design

I think it's related to #31

trnasition-level reward design

I think it's slightly different. In GRPO for RLHF, each trajectory = one generated response. In our setup, each trajectory = multiple responses.

How to run agentic rl with agent lightning in the distributed training environment with multi-nodes and multi-gpus ?

We might need an example for this.

i write a custom agent base on qwen-agent, can i use it in my project?

Hi. I'm not familiar with Qwen Agent. My questions are: 1. Do they work with OpenAI's chat completion API? 2. Are they traceable via observability tools like AgentOps or Weave?

Support for TS/JS based agent frameworks?

I think the answer is yes and no. We do plans to support agents written in other languages, like the undergoing work with Claude Code and Codex. Right now these...

Support for TS/JS based agent frameworks?

We might want a Vercel AI SDK example after finalizing store RESTful API in v0.3.