Phase-1 Experimental: Mistral-Small-3.1-24B-Instruct-2503 Bring-up

[nikileshx]

Ticket

Link to JIRA ticket

Problem description

Added experimental support for the mistralai/Mistral-Small-3.1-24B-Instruct-2503 model.

Changes Introduced

• Integrated support for Mistral-Small-3.1-24B-Instruct-2503.
• Updated model_config.py to handle the new model configuration.
• Extended load_checkpoints.py to enable weight loading for the model.
• Added new submodules to support the model’s vision component.
• Reused the existing text model pipeline to integrate with Mistral-Small-3.1-24B-Instruct-2503

Notes

• No code from tt-transformers was reused for the vision model bring-up.
• All files under experimental were purpose-built or re-written specifically for Mistral-Small-3.1-24B-Instruct-2503.

Checklist

• All post commit CI passes
• Blackhole Post commit CI with demo tests passes (if applicable)
• Model regression CI passes (if applicable)
• Device performance regression CI passes (if applicable)
• (For models and ops writers) Single-card demo tests CI passes (if applicable) See recommended dev flow.
• Galaxy quick CI passes (if applicable)
• TG demo tests, for Llama CI passes, if applicable, because of current Llama work
• (For runtime and ops writers) T3000 unit tests CI passes (if applicable, since this is run on push to main)
• (For models and ops writers) T3000 demo tests CI passes (if applicable, since this is required for release)
• (For models and ops writers) Single-card demo tests CI passes (if applicable, since this is required for release)
• New/Existing tests provide coverage for changes