Isaac Fino comments

Results 9 comments of


                                            Isaac Fino

filter out empty triplets when batching

> # Before fixing > > the server side reports the above error and exits. > > The client side reports error and hangs after server exiting: > > >...

filter out empty triplets when batching

> To warn the prompt length of agent client, we need to intialize a tokenizer to precompute number of tokens during enqueuing a task ([queue_task](https://github.com/microsoft/agent-lightning/blob/d00cc3e4aab227207520c080e30003c5597a988c/agentlightning/server.py#L301)). Then we can add a...

[Over-long prompt/response length] cannot reshape tensor of 0 elements into shape [1, 0, -1, 128] because the unspecified dimension size -1 can be any value and is ambiguous

It would be greatly appreciated if there were an automated process to convert empty or erroneous outputs to prevent tensor shape issues.

[Over-long prompt/response length] cannot reshape tensor of 0 elements into shape [1, 0, -1, 128] because the unspecified dimension size -1 can be any value and is ambiguous

Though after clamping the length of the prompt and strictly controlling the length of LLM, this issue can be avoided.

[Over-long prompt/response length] cannot reshape tensor of 0 elements into shape [1, 0, -1, 128] because the unspecified dimension size -1 can be any value and is ambiguous

# More information. If the agent is a chain of tools and with multi turn responses, this issue happens more frequently. I think the key issue is in the async...

[Over-long prompt/response length] cannot reshape tensor of 0 elements into shape [1, 0, -1, 128] because the unspecified dimension size -1 can be any value and is ambiguous

Yes, I can give a detailed config with my training with the agent. It's a really minimal setting for RL training, even cannot be used in real implementation. GPUS are...

[Over-long prompt/response length] cannot reshape tensor of 0 elements into shape [1, 0, -1, 128] because the unspecified dimension size -1 can be any value and is ambiguous

> Hi [@IsaacGHX](https://github.com/IsaacGHX) , can you provide your script for launching agentlightning training server, e.g. [train.sh for calc_x example](https://github.com/microsoft/agent-lightning/blob/main/examples/calc_x/train.sh)? > > ## Issue analysis > According to the error you...

[QUESTION] main.tex 的开头部分 `\documentclass` 中选项的作用是什么? 可否删除?

历史遗留版本太复杂了x，这个实在是没有机会看完，我目前这里的PR的原因是由于将其他会议格式修改到这里导致的和要求不一致；等毕设结束后续有机会再排查具体设置逻辑

暗牧策略问题

已经针对所有的上述问题提供解决方案，详见25/03/29 PR from me