Skip to content

Modify GPT_J to use easydel#439

Open
ctr-pmuruganTT wants to merge 5 commits intomainfrom
pmurugan/implement_easydel_for_gptj
Open

Modify GPT_J to use easydel#439
ctr-pmuruganTT wants to merge 5 commits intomainfrom
pmurugan/implement_easydel_for_gptj

Conversation

@ctr-pmuruganTT
Copy link
Contributor

Problem description

Add EasyDel implementation for JAX Whisper model

What's changed

  • updated requirements.txt with the requirements for this model
  • updated load_model() to use EasyDel model
  • Added gradient_checkpointing=EasyDeLGradientCheckPointers.NONE as default it is taking a bool value and it is resulting in an error as ValueError: Invalid policy type: <class 'bool'>
  • updated load_inputs() to pass mesh as a parameter
  • added load_parameters_partition_spec() and get_input_activations_partition_spec()

Checklist

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant