Closed oplatek closed 4 months ago
Thanks! All the discrepancies were caused by the quick development of text-generation-webui. Their original API got deprecated in favor of OpenAI-like API.
As a hotfix, I linked to the respective commit of text-generation-webui (still using the original API) in the README.
But since than I came to realize that it is better to have a code that everyone can run instead of an exact replica of my experimental code. I will integrate the changes.
gpu-python --gpu_mem 48 server.py --settings settings_local.yaml --listen --listen-host 0.0.0.0 --api --model "mistralai_Mistral-7B-Instruct-v0.1"
@kasnerz I added dependencies and commit your changes.
Feel free to close the request and commit your changes from our cluster, but please add the requirements