From 61d576a13b0fb17f87b2eb5a6893cac74eccc4e9 Mon Sep 17 00:00:00 2001 From: fzyzcjy Date: Mon, 21 Apr 2025 21:32:37 +0800 Subject: [PATCH] more --- python/sglang/srt/models/deepseek_v2.py | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/python/sglang/srt/models/deepseek_v2.py b/python/sglang/srt/models/deepseek_v2.py index 2cd64d5326e..5f39cedd43f 100644 --- a/python/sglang/srt/models/deepseek_v2.py +++ b/python/sglang/srt/models/deepseek_v2.py @@ -312,12 +312,6 @@ def forward_deepep( self, hidden_states: torch.Tensor, forward_mode: ForwardMode ) -> torch.Tensor: shared_output = None - topk_idx = torch.full( - (0, self.top_k), -1, dtype=torch.int, device=hidden_states.device - ) - topk_weights = torch.empty( - (0, self.top_k), dtype=torch.float32, device=hidden_states.device - ) if ( forward_mode is not None and not forward_mode.is_idle() @@ -337,6 +331,13 @@ def forward_deepep( correction_bias=self.correction_bias, routed_scaling_factor=self.routed_scaling_factor, ) + else: + topk_idx = torch.full( + (0, self.top_k), -1, dtype=torch.int, device=hidden_states.device + ) + topk_weights = torch.empty( + (0, self.top_k), dtype=torch.float32, device=hidden_states.device + ) if self.ep_size > 1: # TODO(ch-wan): allow users to set num_max_dispatch_tokens_per_rank value (