Fix integration test expected values for cuda sm_86 (Mar 15 regressions) (#48168)
* [AyaVision] Update ("cuda", 8) expected value in integration test
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [CWM] Add ("cuda", 8) expected logits in sliding window integration test
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [GLM] Update ("cuda", 8) expected text in test_model_9b_eager
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [Helium] Update ("cuda", 8) expected text in test_model_2b
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [HiggsAudioV2] Add ("cuda", 8) expected token tensors in integration tests
Add cuda sm_86 expected values for test_multi_speaker_smart_voice,
test_zero_shot_voice_cloning, test_multi_speaker_voice_cloning, and
test_batched_inference.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [Swin2SR, SwiftFormer] Update stale expected values for cuda sm_86
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [Mistral] Update stale ("cuda", 8) expected logit value (index 13) for cuda sm_86
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [Voxtral, VoxtralRealtime] Add ("cuda", 8) expected outputs for cuda sm_86
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [Llama, SeedOss] Add ("cuda", 8) expected values for cuda sm_86
- LlamaIntegrationTest::test_llama_3_1_hard: add ("cuda", 8) entry;
sm_86 generates "radical social and political" vs older hw "radical
political and social"
- SeedOssIntegrationTest::test_model_36b_eager/sdpa: convert to
Expectations; sm_86 generates different second-prompt continuation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [seed_oss] Fix fallback Expectations key from ("cuda", None) to (None, None)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* [higgs_audio_v2] Fix missing comma between batch tensors in test_batched_inference
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>