1.fix the bug of draft model with ep 2.fix sampler bug (#4589)

2025-12-24 13:28:13 +08:00 · 2025-10-27 17:47:34 +08:00
parent 8aab4e367f
commit 25a983ba9c
2 changed files with 12 additions and 2 deletions
--- a/fastdeploy/model_executor/layers/sample/sampler.py
+++ b/fastdeploy/model_executor/layers/sample/sampler.py
@@ -567,6 +567,7 @@ class SpeculativeSampler(nn.Layer):
        max_model_len: int,
        share_inputs: List[paddle.Tensor],
        accept_all_drafts: bool = False,
+        reject_all_drafts: bool = False,
    ) -> paddle.Tensor:
        """ """

@@ -622,7 +623,7 @@ class SpeculativeSampler(nn.Layer):
            max_model_len,
            self.speculative_verify_window,
            True,  # enable_topp
-            self.speculative_benchmark_mode,
+            (self.speculative_benchmark_mode or reject_all_drafts),
            accept_all_drafts,
        )