Add test for Openai GPT OSS model

[meenakshiramanathan1]

Model card: https://huggingface.co/blog/welcome-openai-gpt-oss
Model requires transformer version upgrade of 4.55.0 , post which it fails with RuntimeError: Using MXFP4 quantized models requires a GPU. The model uses MXFP4 quantization, which requires a GPU to run. But since we’re planning to use the model in BF16 or FP32 and don’t actually need quantization, quantization_config has been removed from model config before loading model.

After this change, the model loads without the MXFP4 error, but we encounter a warning: Some weights of GptOssForCausalLM were not initialized from the model checkpoint at openai/gpt-oss-20b and are newly initialized, which suggests that the model weights are not being loaded correctly—likely due to the architecture/config mismatch caused by altering the original configuration. This test is intended to verify whether the model's architecture is compatible with Forge compilation, and to determine how far the compilation process progresses.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 50.36%. Comparing base (3486037) to head (297af22).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2749   +/-   ##
=======================================
  Coverage   50.36%   50.36%           
=======================================
  Files         171      171           
  Lines       13042    13042           
=======================================
  Hits         6569     6569           
  Misses       6473     6473           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.