Add Experimental Support for Gemma-3-4b-it

[MohammedTaherMcW]

Ticket

Link to Github Issue

Problem description

Enable Support for Gemma-3-4b-it Model.

What's changed

Added support for gemma-3-4b-it model
Updated model_config.py to support gemma-3-4b-it , including end-of-sequence (EoS) token handling.
Updated load_checkpoints.py to support gemma-3-4b-it weight loading.
Modified apply_scaling logic to handle both LLaMA and gemma-3-4b-it 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

[arginugaTT]

Hi @MohammedTaherMcW

Please add your accuracy tests symlinked to tests/nightly/single_card/model_name and add the entry in .github/workflows/fast-dispatch-full-regressions-and-models-impl.yaml

create an entry for code owners of this model as pavle petrovic: .github/CODEOWNERS

then run the frequent model test workflow and attach to the ticket.

For reference: https://github.com/tenstorrent/tt-metal/pull/20690/files

[jschuhmacher]

For visibility, this was merged as tenstorrent#26924