[SPEC_V2] Fix Acclen drop when enabling DP Attention for Spec-Overlap (#16310)
This commit is contained in:
@@ -491,7 +491,7 @@ class EagleDraftWorker(BaseDraftWorker):
|
|||||||
draft_input = EagleDraftInput(
|
draft_input = EagleDraftInput(
|
||||||
hidden_states=batch_result.logits_output.hidden_states,
|
hidden_states=batch_result.logits_output.hidden_states,
|
||||||
num_tokens_per_batch=self.speculative_num_steps + 1,
|
num_tokens_per_batch=self.speculative_num_steps + 1,
|
||||||
num_tokens_for_logprob_per_batch=1,
|
num_tokens_for_logprob_per_batch=self.speculative_num_steps + 1,
|
||||||
)
|
)
|
||||||
select_index = (
|
select_index = (
|
||||||
torch.arange(len(batch.seq_lens), device=self.device)
|
torch.arange(len(batch.seq_lens), device=self.device)
|
||||||
|
|||||||
Reference in New Issue
Block a user