Add experimental model bringup for google/gemma-3-1b-it

[MohammedTaherMcW]

Ticket

Link to Github Issue

Problem description

Enable Support for Gemma-3-1b-it Model.

What's changed

• Added support for gemma-3-1b-it model
• Updated model_config.py to support datatype (BF16) in gemma-3-1b-it , including end-of-sequence (EoS) token handling and loading via Gemma3CausalLM.
• Updated load_checkpoints.py to support gemma-3-1b-it weight loading.
• Modified apply_scaling logic to handle both LLaMA and gemma-3-1b-it model.
• Added Compute Kernel Config of HiFi4 and Fp32 support.

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

[willwray]

Thanks, reviewing.
To review the first commit, it would be useful to know the origin or 'provenance' of the new files in order to compare for TR-specific changes. We discussed this in last week's call, and it sounded like there is no clean-slate origin, but if you can say where the work started then that will help.

[willwray]

Thanks Mohammed,

I rebased on a new branch and submitted to TT with no changes

tenstorrent#26438
tenstorrent#26439
tenstorrent#26440
tenstorrent#26441