[Bugfix] fix greedy temperature detection (#5417)

### What this PR does / why we need it?
fix greedy temperature detection from
https://github.com/vllm-project/vllm/pull/27077

- vLLM version: release/v0.13.0
- vLLM main:
https://github.com/vllm-project/vllm/commit/81786c87748b0177111dfdc07af5351d8389baa1
---------
Signed-off-by: realliujiaxu <realliujiaxu@163.com>
This commit is contained in:
realliujiaxu authored and GitHub committed 2025-12-27 17:04:10 +08:00
1 parent eab306b09c
commit 2add3dc3e0
1 file changed
+2 -2
+2 -2
View File
@@ -4,12 +4,12 @@ from typing import Optional
import torch
from vllm.triton_utils import HAS_TRITON, tl, triton
from vllm.v1.sample.metadata import SamplingMetadata
from vllm.v1.sample.rejection_sampler import generate_uniform_probs
from vllm.v1.sample.rejection_sampler import (GREEDY_TEMPERATURE,
generate_uniform_probs)
from vllm_ascend.sample.sampler import apply_top_k_top_p
PLACEHOLDER_TOKEN_ID = -1
GREEDY_TEMPERATURE = -1
# Maximum number of speculative draft tokens allowed per request in a single
# step. This value is chosen to be large enough to handle typical use cases.
MAX_SPEC_LEN = 32