Sai Madhusudan Gunda

Results 6 comments of Sai Madhusudan Gunda

The reason for doing this is using the concurrency , multi gpu , multi node , batching properties of vllm library

It is nice if we can add for vllm library then any model can be used #670

Are you sure about it ? ` if self.check_line_spans(lines): page_lines[page_id] = lines # Changes to the page object are not reflected in the dictionary output refs = page.get("refs") if refs...

Basically I am not working with Devanagiri script but with something else of Sanskrit like Modi I wanted to how how can I write a code which will finetune surya...

May be something like a callback function