From fdd5bd49fcc332ea9fba2e3c2d9a21f355eb6605 Mon Sep 17 00:00:00 2001 From: dongfengy <99041270+dongfengy@users.noreply.github.com> Date: Mon, 8 Sep 2025 13:59:28 -0700 Subject: [PATCH] [https://nvbugs/5481080][fix] Fix GPTOSS W4A16 reference (#7323) Signed-off-by: Dongfeng Yu --- tests/integration/defs/accuracy/references/gsm8k.yaml | 2 ++ tests/integration/test_lists/waives.txt | 1 - 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/tests/integration/defs/accuracy/references/gsm8k.yaml b/tests/integration/defs/accuracy/references/gsm8k.yaml index ca5f921df3..f38f2927e4 100644 --- a/tests/integration/defs/accuracy/references/gsm8k.yaml +++ b/tests/integration/defs/accuracy/references/gsm8k.yaml @@ -204,5 +204,7 @@ GPT-OSS/MXFP4: accuracy: 90.3 - quant_algo: W4A8_MXFP4_FP8 accuracy: 90.3 + - quant_algo: W4A16_MXFP4 + accuracy: 90.3 LGAI-EXAONE/EXAONE-4.0-32B: - accuracy: 88.36 diff --git a/tests/integration/test_lists/waives.txt b/tests/integration/test_lists/waives.txt index 18968eb4a0..b3416ab543 100644 --- a/tests/integration/test_lists/waives.txt +++ b/tests/integration/test_lists/waives.txt @@ -329,7 +329,6 @@ accuracy/test_cli_flow.py::TestLongAlpaca7B::test_auto_dtype SKIP (https://nvbug accuracy/test_llm_api.py::TestPhi4MiniInstruct::test_fp8 SKIP (https://nvbugs/5465143) accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_fp8_blockscale[throughput] SKIP (https://nvbugs/5471106) accuracy/test_llm_api_pytorch.py::TestEXAONE4::test_auto_dtype SKIP (https://nvbugs/5481090) -accuracy/test_llm_api_pytorch.py::TestGPTOSS::test_w4_1gpu[True-True-cutlass] SKIP (https://nvbugs/5481080) test_e2e.py::test_ptp_quickstart_advanced_8gpus_chunked_prefill_sq_22k[Llama-4-Maverick-17B-128E-Instruct-FP8-llama4-models/nvidia/Llama-4-Maverick-17B-128E-Instruct-FP8-False] SKIP (https://nvbugs/5481094) test_e2e.py::test_ptp_quickstart_advanced_8gpus_chunked_prefill_sq_22k[Llama-4-Maverick-17B-128E-Instruct-FP8-llama4-models/nvidia/Llama-4-Maverick-17B-128E-Instruct-FP8-True] SKIP (https://nvbugs/5481094) test_e2e.py::test_ptp_quickstart_advanced_8gpus_chunked_prefill_sq_22k[Llama-4-Scout-17B-16E-Instruct-FP8-llama4-models/Llama-4-Scout-17B-16E-Instruct-FP8-True] SKIP (https://nvbugs/5481094)