From 15c8c240b514cd19e7a87a99b61468b874889d5f Mon Sep 17 00:00:00 2001
From: bukejiyu <52310069+bukejiyu@users.noreply.github.com>
Date: Tue, 15 Jul 2025 00:39:12 +0800
Subject: [PATCH] [vl] Use top_k from config.json (#2831)

---
 fastdeploy/worker/vl_gpu_model_runner.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/fastdeploy/worker/vl_gpu_model_runner.py b/fastdeploy/worker/vl_gpu_model_runner.py
index 82af454b5..778ca8424 100644
--- a/fastdeploy/worker/vl_gpu_model_runner.py
+++ b/fastdeploy/worker/vl_gpu_model_runner.py
@@ -235,8 +235,6 @@ class GPUVLModelRunner(VLModelRunnerBase):
         fd_config.model_config.think_end_id = tokenizer.get_vocab()["</think>"]
         fd_config.model_config.max_text_id = fd_config.model_config.im_patch_id
         fd_config.model_config.sequence_parallel = False
-        # TODO(YuanRisheng) The moe_k in develop is fixed to 8, need to be changed according to json config
-        fd_config.model_config.moe_k = 8
         self.fd_config = fd_config
         self.model_cfg = self.fd_config.model_config
         self.image_preprocess = self._init_image_preprocess(