Skip to content

Mismatching default value of Llama4TextConfig attn_temperature_tuning between official llama code #37479

@gmlwns2000

Description

@gmlwns2000

System Info

In transfomers library, the default value of attn_temperature_tuning in Llama4TextConfig, is 4.

attn_temperature_tuning (`int`, *optional*, defaults to 4): TODO

However, I think this value should be boolean because it is used as a condition flag in the forward pass.

if self.attn_temperature_tuning and not self.use_rope:

Moreover, in the official implementation, that value is boolean, which is defaulted to False. (Offical Config File)

Who can help?

@ArthurZucker

Information

  • The official example scripts
  • My own modified scripts

Tasks

  • An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
  • My own task or dataset (give details below)

Reproduction

Any Llama4TextConfig instances.

Expected behavior

The default value of config.attn_temperature_tuning should be True or False depending on input sequence length.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions