You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have pre-trained and fine-tuned a MPT 350m model using llm-foundry. Upon conversion of the composer checkpoint to a HF one, I tried hosting it using vLLM. I find that vLLM asserts the attn_config.alibi flag is set to in the config.json file. However, this flag is set as False in my model's config.json file.
Is there a wayaround to host my model. Any solution or directions will be of great help. Thanks!
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
I have pre-trained and fine-tuned a MPT 350m model using llm-foundry. Upon conversion of the composer checkpoint to a HF one, I tried hosting it using vLLM. I find that vLLM asserts the
attn_config.alibi
flag is set to in theconfig.json
file. However, this flag is set asFalse
in my model'sconfig.json
file.Is there a wayaround to host my model. Any solution or directions will be of great help. Thanks!
The
config.json
file is as below -Beta Was this translation helpful? Give feedback.
All reactions