From d86e3f0172e29531215ce0ad4c93785da23dc43f Mon Sep 17 00:00:00 2001 From: Bill Nell Date: Wed, 28 May 2025 23:38:19 +0000 Subject: [PATCH] lint Signed-off-by: Bill Nell --- .../layers/fused_moe/fused_batched_moe.py | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/vllm/model_executor/layers/fused_moe/fused_batched_moe.py b/vllm/model_executor/layers/fused_moe/fused_batched_moe.py index bd0489f953ea0..7d42894ffe1a1 100644 --- a/vllm/model_executor/layers/fused_moe/fused_batched_moe.py +++ b/vllm/model_executor/layers/fused_moe/fused_batched_moe.py @@ -614,13 +614,10 @@ class BatchedExperts(mk.FusedMoEPermuteExpertsUnpermute): else: num = int(expert_num_tokens[expert].item()) tmp = _resize_cache(workspace2, (num, N)) - if self.use_fp8_w8a8: - assert False # TBD - else: - input = hidden_states[expert, :num, :] @ w1[expert].transpose( - 0, 1) - self.activation(activation, tmp, input) - out[expert, :num, :] = tmp @ w2[expert].transpose(0, 1) + assert not self.use_fp8_w8a8 + input = hidden_states[expert, :num, :] @ w1[expert].transpose(0, 1) + self.activation(activation, tmp, input) + out[expert, :num, :] = tmp @ w2[expert].transpose(0, 1) return out