From dedea1dfabceb73e9497362c3c1d4f2e0e13fb6e Mon Sep 17 00:00:00 2001 From: CSY-ModelCloud Date: Fri, 22 Aug 2025 16:55:09 +0800 Subject: [PATCH 1/2] [CI] use new docker without pip uv --- .github/workflows/unit_tests.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/.github/workflows/unit_tests.yml b/.github/workflows/unit_tests.yml index 11fb0143d..adf0d4f30 100644 --- a/.github/workflows/unit_tests.yml +++ b/.github/workflows/unit_tests.yml @@ -205,7 +205,7 @@ jobs: - list-test-files if: needs.list-test-files.outputs.torch-files != '[]' || needs.list-test-files.outputs.transformers-files != '[]' container: - image: ${{ needs.check-vm.outputs.ip }}:5000/nvidia/cuda:${{ needs.check-vm.outputs.cuda_version }}-ubuntu22.04 + image: ${{ needs.check-vm.outputs.ip }}:5000/nvidia/cuda:${{ needs.check-vm.outputs.cuda_version }}-ubuntu22.04_0822 options: --device /dev/dri --ipc=host --runtime=nvidia --gpus all volumes: - /dev/dri/by-path:/dev/dri/by-path @@ -327,7 +327,7 @@ jobs: - check-vm runs-on: [ self-hosted, xeon5 ] container: - image: ${{ needs.check-vm.outputs.ip }}:5000/nvidia/cuda:${{ needs.check-vm.outputs.cuda_version }}-ubuntu22.04 + image: ${{ needs.check-vm.outputs.ip }}:5000/nvidia/cuda:${{ needs.check-vm.outputs.cuda_version }}-ubuntu22.04_0822 options: --device /dev/dri --ipc=host --runtime=nvidia --gpus all volumes: - /dev/dri/by-path:/dev/dri/by-path From c5505d612a0f2031e0117bd620bef64be6469318 Mon Sep 17 00:00:00 2001 From: CSY-ModelCloud Date: Fri, 22 Aug 2025 17:46:29 +0800 Subject: [PATCH 2/2] fix ipex error --- tests/test_packable.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/tests/test_packable.py b/tests/test_packable.py index d1590cfee..7d1ed90b1 100644 --- a/tests/test_packable.py +++ b/tests/test_packable.py @@ -8,7 +8,7 @@ from gptqmodel.nn_modules.qlinear.exllama import ExllamaQuantLinear # noqa: E402 from gptqmodel.nn_modules.qlinear.exllama_eora import ExllamaEoraQuantLinear from gptqmodel.nn_modules.qlinear.exllamav2 import ExllamaV2QuantLinear # noqa: E402 -from gptqmodel.nn_modules.qlinear.ipex import IPEXQuantLinear # noqa: E402 +from gptqmodel.nn_modules.qlinear.ipex import IPEXQuantLinear, HAS_IPEX # noqa: E402 from gptqmodel.nn_modules.qlinear.marlin import MarlinQuantLinear # noqa: E402 from gptqmodel.nn_modules.qlinear.torch import TorchQuantLinear # noqa: E402 from gptqmodel.nn_modules.qlinear.tritonv2 import TritonV2QuantLinear # noqa: E402 @@ -59,6 +59,8 @@ def setUpClass(cls): ] ) def test_post_init(self, backend: BACKEND, equal: Dict[str, bool]): + if backend == BACKEND.IPEX and not HAS_IPEX: + self.skipTest("IPEX is not available") model = GPTQModel.load(self.model_id, backend=backend, device_map="auto") model = convert_gptq_v2_to_v1_format(model, model.quantize_config, self.QLINEAR_DICT[backend])