huggingface / optimum-quanto

A pytorch quantization backend for optimum
Apache License 2.0
833 stars 61 forks source link

[tests] enable `test_weight_qbits_tensor_linear_cuda` on xpu devices #345

Closed faaany closed 1 week ago

faaany commented 3 weeks ago

What does this PR do?

This is a follow-up PR of #344. This should be merged after #344.

Below are the test results:

PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-32-1-bf16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-32-2-fp16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-32-2-bf16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-48-1-fp16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-48-1-bf16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-48-2-fp16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-48-2-bf16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-64-1-fp16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-64-1-bf16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-64-2-fp16]
PASSED test/tensor/weights/test_weight_qbits_tensor_dispatch.py::test_weight_qbits_tensor_linear_gpu[no-bias-4096-16384-64-2-bf16]
================================================ 288 passed, 161 deselected in 4.56s =================================================

Before submitting

faaany commented 1 week ago

Thanks for the rebase and merge!