From 607fadea787473fbe86ac4011c1b7bd568875c7e Mon Sep 17 00:00:00 2001 From: Kai Xu Date: Fri, 20 Feb 2026 22:57:45 -0800 Subject: [PATCH] Fix the example test Signed-off-by: Kai Xu --- tests/examples/llm_eval/test_llm_eval.py | 17 ----------------- 1 file changed, 17 deletions(-) diff --git a/tests/examples/llm_eval/test_llm_eval.py b/tests/examples/llm_eval/test_llm_eval.py index 88d29dedc..0abf78b53 100644 --- a/tests/examples/llm_eval/test_llm_eval.py +++ b/tests/examples/llm_eval/test_llm_eval.py @@ -36,20 +36,3 @@ def test_llama_eval_fp8(): finally: # Force kill llm-serve if it's still running subprocess.run(["pkill", "-f", "llm-serve"], check=False) - - -def test_llama_eval_sparse_attention(tiny_llama_path): - """Test sparse attention with llm_eval integration.""" - try: - # Test with default sparse attention config (no quantization) - run_llm_ptq_command( - model=tiny_llama_path, - quant="none", # No quantization, only sparse attention - tasks="lm_eval", - lm_eval_tasks="hellaswag", - lm_eval_limit=0.05, # Small limit for fast test - sparse_cfg="SKIP_SOFTMAX_DEFAULT", - batch=4, - ) - finally: - subprocess.run(["pkill", "-f", "llm-serve"], check=False)