Skip to content

Commit

Permalink
functional test fix
Browse files Browse the repository at this point in the history
  • Loading branch information
isanghao committed Nov 13, 2024
1 parent 77a74c4 commit ee3b6e8
Show file tree
Hide file tree
Showing 2 changed files with 6 additions and 2 deletions.
4 changes: 4 additions & 0 deletions src/plugins/intel_gpu/src/plugin/transformations_pipeline.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -919,6 +919,10 @@ void TransformationsPipeline::apply(std::shared_ptr<ov::Model> func) {
GPU_DEBUG_TRACE << root->get_friendly_name() << " : dynamic quantization is turned off because weight may not representable in 8 bit" << std::endl;
return true;
}

// Currently only per-token quantization is supported for onednn path
if (dynamic_quantization_group_size != UINT64_MAX)
return true;
return false;
});
manager.register_pass<ov::intel_gpu::DynamicQuantizeFullyConnected>(dynamic_quantization_group_size);
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -404,10 +404,10 @@ INSTANTIATE_TEST_SUITE_P(MatMulCompressedWeights_corner_cases_big,
// per_tensor_zp=0 is not supported
// transpose_weights is not supported
// weight precision u4 is only supported
const std::vector<uint64_t> group_size = {32, 128, 256, UINT64_MAX};
const std::vector<uint64_t> group_size = {32, 128, UINT64_MAX};
INSTANTIATE_TEST_SUITE_P(smoke_MatMulCompressedWeights_dyn_quan,
MatmulWeightsDecompression,
::testing::Combine(::testing::Values(ShapeParams{{{-1, -1, 4096}, {{1, 1, 4096}}}, {1, 4096, 4096}}), // shape
::testing::Combine(::testing::Values(ShapeParams{{{-1, -1, 4096}, {{1, 1, 4096}}}, {4096, 4096}, 128}), // shape
::testing::Values(ov::element::u4),
::testing::Values(ov::element::f16),
::testing::Values(false),
Expand Down

0 comments on commit ee3b6e8

Please sign in to comment.