[New feature] Need "decoder_start_token" in inferencing #126

Open

Labels

opened

Need to add a new parameter in inferencing configuration that allows the user to assign a "decoder_start_token" or "decoder_start_token_id" for model inferencing.

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

Metadata

Assignees

No one assigned

Labels

Type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests