From 33322f2625def6501b40c7317db24d8be369e909 Mon Sep 17 00:00:00 2001 From: "Jin, Youzhi" Date: Tue, 22 Sep 2026 14:54:29 +0000 Subject: [PATCH 1/2] Fix AttributeError for pytest mismatch Signed-off-by: Jin, Youzhi --- tests/unit/checkpoint/test_convert_checkpoint.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/unit/checkpoint/test_convert_checkpoint.py b/tests/unit/checkpoint/test_convert_checkpoint.py index 5d3856e95f59..9ac0b70d6546 100644 --- a/tests/unit/checkpoint/test_convert_checkpoint.py +++ b/tests/unit/checkpoint/test_convert_checkpoint.py @@ -160,6 +160,6 @@ def test_convert_zero_checkpoint_to_fp32_state_dict(self, tmpdir): assert tensor.dtype == torch.bfloat16 torch.testing.assert_close(tensor.float(), state_dict[name], rtol=5e-3, atol=5e-3) - fp32_size = (fp32_save_dir / 'pytorch_model.bin').stat().st_size - bf16_size = (bf16_save_dir / 'pytorch_model.bin').stat().st_size + fp32_size = (fp32_save_dir / 'pytorch_model.bin').stat().size + bf16_size = (bf16_save_dir / 'pytorch_model.bin').stat().size assert bf16_size < fp32_size * 0.6 From 0b1d39b176f1d27f6d58c6198b1480d879e737c7 Mon Sep 17 00:00:00 2001 From: "Jin, Youzhi" Date: Wed, 23 Sep 2026 03:37:33 +0000 Subject: [PATCH 2/2] Skip half int8 quantization on CPU Signed-off-by: Jin, Youzhi --- tests/unit/inference/quantization/test_intX_quantization.py | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/unit/inference/quantization/test_intX_quantization.py b/tests/unit/inference/quantization/test_intX_quantization.py index 18a491123b84..29bbfdf95936 100644 --- a/tests/unit/inference/quantization/test_intX_quantization.py +++ b/tests/unit/inference/quantization/test_intX_quantization.py @@ -378,6 +378,7 @@ def test_float_int8_quantization(self): reset_random() quantization_test_helper(torch.float32, 8) + @pytest.mark.skipif(device == 'cpu', reason='CPU does support FP16 GEMM') def test_half_int8_quantization(self): reset_random() quantization_test_helper(torch.float16, 8)