hotfix - Revert vllm/attention/layer.py changes from 0f8cafe - fix torch.compile recompilations #518

Job	Run time
CodeQL Scan	3m 2s
Dependency Review	4s
Calculate Tests To Trigger	6s
Test / gsm8k_small_g3_tp1	8m 39s
Test / gsm8k_small_g3_tp2	1h 0m 44s
Test / gsm8k_small_g2_tp1	8m 13s
Test / gsm8k_small_g2_tp2	11m 39s
Test / gsm8k_large_g3_tp2	8m 32s
Test / gsm8k_large_g2_tp4	7m 10s
Test / gsm8k_small_g3_tp1_fp8	8m 28s
Test / gsm8k_small_g3_tp2_fp8	8m 14s
Test / gsm8k_small_g3_tp1_mss	4m 18s
Test / gsm8k_small_g2_tp1_mss	9m 37s
Test / gsm8k_small_g3_tp2_mss	7m 58s
Test / gsm8k_small_g2_tp2_mss	9m 24s
Test / gsm8k_small_g2_tp1_spec_decode	9m 39s
Test / gsm8k_small_g2_tp1_mlp_spec_decode	4m 46s
Test / gsm8k_small_g2_tp1_medusa_spec_decode	4m 24s
Test / gsm8k_small_g2_tp1_eagle_spec_decode	5m 9s
Test / test_llama_lora	8m 15s
Test / test_multilora	8m 36s
Test / test_long_context	10m 3s
	3h 27m 0s

Provide feedback