Qwen 2.5 VL 7B - Phase 1 - experimental stage

[MohammedTaherMcW]

Ticket

Link to JIRA Issue

Problem description

Enabled Support for Qwen2.5-VL-7B-Instruct Model.

What's changed

• Added support for Qwen2.5-VL-7B-Instruct
• Updated model_config.py to support Qwen2.5-VL-7B-Instruct.
• Updated load_checkpoints.py to support Qwen2.5-VL-7B-Instruct weight loading.
• Added new submodules to support the Qwen-VL 2.5 vision model.
• Modified the text MLP to avoid L1 buffer issue for Qwen

Note

No code from tt-transformers has been re-used in experimental; all files in experimental are specifically tailored for the Qwen 2.5 VL model.

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