[Bugfix] bugfix for moe_mlp in vllm-ascend/v0.11.0-dev (#4885)
### What this PR does / why we need it? This PR fixes a bug in the moe_mlp module by correcting the arguments passed to the torch_npu.npu_dequant_swiglu_quant function.It properly converts group_list from a cumulative sum to counts for the group_index parameter. ### Does this PR introduce _any_ user-facing change? No - vLLM version: v0.12.0 - vLLM main: https://github.com/vllm-project/vllm/main --------- Signed-off-by: tanqingshan (A) <50050625@china.huawei.com> Signed-off-by: tanqingshan (A) <50050625@china.huawei.com> Co-authored-by: tanqingshan (A) <50050625@china.huawei.com> Co-authored-by: Mercykid-bash <ruanche0218@gmail.com>
This commit is contained in:
@@ -130,7 +130,7 @@ class TestTokenDispatcherWithMC2(TestBase):
|
||||
self.dispatcher.need_extra_args = True
|
||||
self.dispatcher.enable_dispatch_v2 = True
|
||||
self.dispatcher.output = torch.randint(0, 8, (10, 1))
|
||||
|
||||
self.dispatcher.moe_expert_num = len(self.dispatcher.expert_map)
|
||||
kwargs = self.dispatcher.get_combine_mc_kwargs(hidden_states)
|
||||
self.assertIn("tp_send_counts", kwargs)
|
||||
|
||||
@@ -148,6 +148,7 @@ class TestTokenDispatcherWithMC2(TestBase):
|
||||
self.dispatcher.enable_dispatch_v2 = True
|
||||
self.dispatcher.swiglu_out_scale = torch.randint(0, 8, (10, 1))
|
||||
self.dispatcher.output = torch.randint(0, 8, (10, 1))
|
||||
self.dispatcher.moe_expert_num = len(self.dispatcher.expert_map)
|
||||
self.hidden_states = torch.randn(10, 128)
|
||||
|
||||
with patch("torch_npu.npu_moe_distribute_combine_v2",
|
||||
|
||||
Reference in New Issue
Block a user