Thanks to visit codestin.com
Credit goes to github.com

Skip to content

How torchax handle tied_embedding? #68

@carlesoctav

Description

@carlesoctav

This is just my Observation: loading weights from safetensors (or maybe getting the weights from state_dict()) results in two weights even when the model uses tied_embedding, model.embed_tokens and lm_head, If we use these two weights in the forward call, we'll chagne both weights instead of just one, no?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions