A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 8, 08:45 UTC
[Bugfix] Delete unused redundant code in Kimi-K2.5 (#34427)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn> Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
LoganJane committed
Feb 13, 2026 at 10:18 UTC
62788f99a4d0e483a6e9114e6708489b44b51a78
1 file changed
-5
vllm/model_executor/models/kimi_k25.py
-5
@@ -11,7 +11,6 @@ This module defines:
11
- KimiK25ForConditionalGeneration: Main model class
12
"""
13
14
-import copy
14
from collections.abc import Iterable, Mapping, Sequence
15
from dataclasses import dataclass
16
from typing import Annotated, Any, Literal
@@ -378,10 +377,6 @@ class KimiK25ForConditionalGeneration(
377
)
378
379
self.quant_config = quant_config
381
- sub_vllm_config = copy.deepcopy(vllm_config)
382
- sub_vllm_config.model_config.hf_config = (
383
- sub_vllm_config.model_config.hf_config.text_config
384
- )
380
with self._mark_language_model(vllm_config):
381
self.language_model = init_vllm_registered_model(
382
vllm_config=vllm_config,