We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent c582024 commit a6f8814Copy full SHA for a6f8814
vllm_ascend/torchair/models/torchair_deepseek_mtp.py
@@ -150,8 +150,9 @@ def forward(
150
inputs_embeds: Optional[torch.Tensor] = None,
151
spec_step_idx: int = 0,
152
) -> torch.Tensor:
153
- if inputs_embeds is None:
+ if inputs_embeds is None and self.embed_tokens is not None:
154
inputs_embeds = self.embed_tokens(input_ids)
155
+
156
current_step_idx = (spec_step_idx % self.num_mtp_layers)
157
step_kv_cache = kv_caches[
158
current_step_idx] if kv_caches is not None else None
0 commit comments