Add Llama Scope SAEs & improvements to evaluating ce scores.

[Hzfinfdu]

Description

This pull request includes the following updates:

• add entries to Llama Scope SAEs in pretrained_saes.yaml
• implement Llama Scope loaders in pretrained_sae_loaders.py
• implement an option to not replacing bos activations when computing ce_loss_score in eval.py. This is necessary for SAEs trained with context starting with bos but is not trained with bos activations.

Fixes # (issue)

Type of change

Please delete options that are not relevant.

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

Checklist:

• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I have not rewritten tests relating to key interfaces which would affect backward compatibility

You have tested formatting, typing and unit tests (acceptance tests not currently in use)

• I have run make check-ci to check format and linting. (you can run make format to format code if needed.)

Performance Check.

If you have implemented a training change, please indicate precisely how performance changes with respect to the following metrics:

• L0
• CE Loss
• MSE Loss
• Feature Dashboard Interpretability

Please links to wandb dashboards with a control and test group.

[chanind]

This PR has typing errors which are failing both here and now in main

[Hzfinfdu]

Sry for that. Let me check this.

[chanind]

It's OK, we just shouldn't merge PRs until CI passes in future