Nicolas Patry
Nicolas Patry
Last commit is from 3 years ago...
Definitely not actively maintained. `tokenizers` and its node bindings are, but they are in maintenance mode only (until someone wants to help redo the bindings with latest neon)
@Pierrci Are you ok with archiving this ?
Hi, for the whole I/O thing, I want to mention we have 1 document at least for pipelines: https://www.notion.so/huggingface2/Pipelines-Guidelines-1de828d4e56e4adb886253440657e13a (Just so that we can try and share some definitions amongst...
That does seem like a bug. Looking into how to fix that properly.
Would you be willing to open a PR for that ? :D
Also coming in a future release pretty soon some fix for torch==2.0 (bogus warning)
`text-generation-server download-weights MODEL_ID` is probably what you're looking for. You can also add the progressbar back using the env variable `HF_HUB_ENABLE_HF_TRANSFER=0` (but download will be slower)
Perfect. Closing this issue then.
It seems you are doing nothing in parallel, meaning every document is processed independantly. You are loosing on the batching aspect to speed things up here. Try sending requests in...