From d5a8079eb6686e4a058214696e6ed06e01e43a23 Mon Sep 17 00:00:00 2001 From: QI JUN <22017000+QiJune@users.noreply.github.com> Date: Thu, 5 Jun 2025 17:21:07 +0800 Subject: [PATCH] Revert "[infra] Unwaive unittests/_torch" (#4950) --- tests/integration/test_lists/waives.txt | 7 +------ tests/unittest/_torch/speculative/test_eagle3.py | 1 - 2 files changed, 1 insertion(+), 7 deletions(-) diff --git a/tests/integration/test_lists/waives.txt b/tests/integration/test_lists/waives.txt index aaf83d1c0f..52befb2c5c 100644 --- a/tests/integration/test_lists/waives.txt +++ b/tests/integration/test_lists/waives.txt @@ -380,14 +380,9 @@ full:B200/examples/test_qwen.py::test_llm_qwen_7b_multi_gpus_summary[qwen2.5_7b_ accuracy/test_cli_flow.py::TestMixtral8x22B::test_int8_plugin_tp8[renormalize-tensor_parallel] SKIP (https://nvbugs/5273695) examples/test_gpt.py::test_starcoder_fp8_quantization_2gpu[starcoder] SKIP (https://nvbugs/5144931) examples/test_gpt.py::test_starcoder_fp8_quantization_2gpu[starcoderplus] SKIP (https://nvbugs/5144931) +unittest/_torch -k "not (modeling or multi_gpu or auto_deploy)" SKIP (https://nvbugs/5280806) examples/test_whisper.py::test_llm_whisper_general[large-v3-disable_gemm_plugin-disable_attention_plugin-disable_weight_only-float16-nb:1-use_python_runtime] SKIP (https://nvbugs/5244570) unittest/_torch/speculative/test_eagle3.py SKIP (https://nvbugs/5280806) -unittest/_torch/modules/test_fused_moe.py SKIP (https://nvbugspro.nvidia.com/bug/5324229) -unittest/_torch/modules/test_moe_load_balancer.py SKIP (https://nvbugspro.nvidia.com/bug/5324229) -unittest/_torch/speculative/test_ngram.py SKIP (https://nvbugspro.nvidia.com/bug/5324239) -unittest/_torch/test_pytorch_model_engine.py SKIP (https://nvbugspro.nvidia.com/bug/5324248) -unittest/_torch/test_resource_manager.py SKIP (https://nvbugspro.nvidia.com/bug/5324252) -unittest/_torch/thop/test_selective_scan_op.py SKIP (https://nvbugspro.nvidia.com/bug/5324258) triton_server/test_triton_rcca.py::test_mistral_beam_search[rcca_4714407-True-10---False-True-False-0-128-disableDecoupleMode-inflight_fused_batching-disableTrtOverlap--guaranteed_no_evict---1-1-1-False-ensemble] SKIP (https://nvbugs/5240060) triton_server/test_triton.py::test_triton_extensive[triton-extensive] SKIP triton_server/test_triton.py::test_gpt_speculative_decoding[gpt-speculative-decoding] SKIP diff --git a/tests/unittest/_torch/speculative/test_eagle3.py b/tests/unittest/_torch/speculative/test_eagle3.py index 1906f47163..b698c8a389 100644 --- a/tests/unittest/_torch/speculative/test_eagle3.py +++ b/tests/unittest/_torch/speculative/test_eagle3.py @@ -17,7 +17,6 @@ from utils.llm_data import llm_models_root [[True, "TRTLLM"], [False, "TRTLLM"], [True, "FLASHINFER"], [False, "FLASHINFER"]]) def test_llama_eagle3(use_cuda_graph: bool, attn_backend: str): - pytest.skip("Test is hanging") total_mem_gb = torch.cuda.get_device_properties(0).total_memory / 1e9 if total_mem_gb < 35: pytest.skip("Not enough memory to load target + draft model")