Initial Verification Run

[jlarson4]

Description

Running the verify_models.py script on the top 10 downloaded models for each architecture & ensuring compatibility where possible.

• Update the docs to include a filterable table of model information
• Update benchmarks to improve memory efficiency and ensure caches are properly cleared when running on MPS
  • Specifically, added a conserve memory flag, to reduce model load size to 1x in Phase 1, to allow larger models to run
• Add rope_theta handling to loading_from_pretrained for transformers v5
• Fix for missing position_ids in some models
• Updated supported_models and verification_history to include the latest benchmark run results
• Added additional flags and improvements to verify_models.py to ensure more models run correctly

Details

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

Screenshots

New DOCS page for supported TransformerBridge model support:

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