Yuge Zhang
Yuge Zhang
If the rollout returns a Rollout object, I think the auto traced information would be suppressed. https://github.com/microsoft/agent-lightning/blob/d00cc3e4aab227207520c080e30003c5597a988c/agentlightning/litagent.py#L117 There is currently no dummy tracer that does nothing. You may implement one...
It sounds good. One problem though. I recommend using the exact same tokens that the inference engine of the model used in prompt consumption and response generation. Retokenization of the...
More information please.
We rely on existing tracing tools to help us visualize. The support to periodically upload traces to agentops is tracked in #30. It still needs more testing and I believe...
What's your expected behavior here? The server should not return 400 if the prompt is too long?
I think it's related to local model path. How do you use the local model? What's the configuration path etc.?
We might need to maintain a local model example on CI. Putting it into backlog. Stay tuned.
We haven't provided a client SDK in golang. So in order to use it in golang, you need to implement an SDK on your own based on the protocol between...
The core problem here is how to collect the trace data. If you are using Python, we have agentops, along with other tracing tools. I'm not sure whether you can...
In server.py, we already had a `num_claims`. Maybe rename it as `claim_sequence_id` is a good idea? I don't see the API used elsewhere. so it should be safe.