minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[NPU] fix dsv4 mtp condition on NPU graph (#32711)

Browse Source
This commit is contained in:
JiaruiChang5268
2026-07-29 20:13:31 +08:00
committed by GitHub
parent d254ec9ff8
commit ca6e0ff2c8
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/hardware_backend/npu/graph_runner/eagle_draft_extend_npu_graph_runner.py
+1 -1
View File
@@ -36,7 +36,7 @@ class EAGLEDraftExtendNpuGraphRunner(EAGLEDraftExtendCudaGraphRunner):
def _replay_graph(self, shape_key, forward_batch): def _replay_graph(self, shape_key, forward_batch):
hf_config = self.model_runner.model_config.hf_config hf_config = self.model_runner.model_config.hf_config
if not (is_deepseek_dsa(hf_config) and is_deepseek_v4(hf_config)): if not (is_deepseek_dsa(hf_config) or is_deepseek_v4(hf_config)):
seq_lens = forward_batch.seq_lens_cpu.tolist() + [0] * ( seq_lens = forward_batch.seq_lens_cpu.tolist() + [0] * (
self.bs - self.raw_bs self.bs - self.raw_bs
) )
python/sglang/srt/hardware_backend/npu/graph_runner/eagle_draft_npu_graph_runner.py
+1 -1
View File
@@ -88,7 +88,7 @@ class EAGLEDraftNpuGraphRunner(EAGLEDraftCudaGraphRunner):
def _replay_graph(self, shape_key, forward_batch): def _replay_graph(self, shape_key, forward_batch):
hf_config = self.model_runner.model_config.hf_config hf_config = self.model_runner.model_config.hf_config
if not (is_deepseek_dsa(hf_config) and is_deepseek_v4(hf_config)): if not (is_deepseek_dsa(hf_config) or is_deepseek_v4(hf_config)):
seq_lens_for_each_draft_step = [] seq_lens_for_each_draft_step = []
for speculative_step_id in range(self.speculative_num_steps - 1): for speculative_step_id in range(self.speculative_num_steps - 1):
seq_lens_cpu = ( seq_lens_cpu = (