Philipp Schmid

icon indicating a website link www.philschmid.de

icon company @huggingface icon location Nuremberg icon location AI Developer Experience @google-deepmind | prev: Tech Lead at @huggingface , AWS ML Hero

Results 136 comments of Philipp Schmid

Pydantic problem

Can you please share the versions you have installed?

Out of Memory: Cannot reproduce T5-XXL run on 8xA10G.

What versions of the libraries do you use? What sequence length do you use?

Need to provide additional args to InferenceClient

Hello @dylan-stark, Currently there is no way to [customize the `InferenceClient`](https://github.com/philschmid/easyllm/blob/a651e9dc28168441276ab3f9d5b1c3c2765ae735/easyllm/clients/huggingface.py#L165). If you want to can create a PR with a recommendation on how to add those.

Question regarding image modes

Can you please provide more information? Like the error you get, code to reproduce the error etc.

Boto dependency shouldn't be foreced

What error are you getting? can you please share it?

Boto dependency shouldn't be foreced

Let me look at that but. You should be able to install `pip install easyllm[bedrock]`

Boto dependency shouldn't be foreced

I pushed an updated version.

Precision Issue

What llama model size are you using?

Precision Issue

Did you make any other change to the code than the model id? What GPU are you using?

(Chat)Completion objects cannot generate diverse outputs

This most likely due to the fact the inference API caching the requests.