Hugging Face's logo

Salesforce
/
codegen-350M-multi

Text Generation
Transformers
PyTorch
codegen
Inference Endpoints
Model card Files Community
6

KeyError: 'codegen'

#1
by EnriqueD - opened
Discussion
EnriqueD

Hi all,

I am running the example from the "how to use" section of the model card. AutoModelForCausalLM.from_pretrained('Salesforce/codegen-350M-multi') gives a KeyError for 'codgen', the model_type in the config. This seems to imply that I am missing a model file or other dependency, does anyone know how to fix this? Thanks!

I am running transformers v4.19.4 btw.

rooa
 Salesforce org

Hi @EnriqueD ,

Thanks for your interest!!

This is because CodeGen hasn't been merged into the main branch yet (pull request). It should be ready on main branch very soon.
If you'd like to give the model a try, please take a look at the original repo here in the mean time.

I will ping here when the pull request is merged.

EnriqueD

Thanks for the prompt and helpful reply!
Looking forward to your ping!

rooa
 Salesforce org

Hi @EnriqueD ,

Good news, the pull request has just been merged! Could you please install transformers from the latest main branch and try again? Thanks!

EnriqueD

Hi @rooa ,
I just tested it out and it works fine now, thank you!

Sign up or log in to comment