FastChat
FastChat copied to clipboard
Fix port in worker-address in vllm_worker for #2467
When just change --port , it has no error but it does not work well when run sevural models in one machine.
And this does not change the behavior when explicitly specify the -worker-address.
Why are these changes needed?
When running more than one model with one controller in one machine, changing --port is a natural behavior.
This a more frequent case should be supported.
Related issue number
Open #2467"
Checks
- [x] I've run
format.shto lint the changes in this PR. - [x] I've included any doc changes needed.
- [x] I've made sure the relevant tests are passing (if applicable).
@merrymercy can help to review this?