Move mem_fraction_static adjustment for multimodal models to server_args.py & Fix session control & Other cleanups (#7748)

2025-07-04 16:33:33 -07:00
parent 975a5ec69c
commit 14229ccf8f
16 changed files with 339 additions and 137 deletions
--- a/python/sglang/srt/models/qwen3.py
+++ b/python/sglang/srt/models/qwen3.py
@@ -11,8 +11,6 @@ from sglang.srt.distributed import (
    get_pp_group,
    get_tensor_model_parallel_rank,
    get_tensor_model_parallel_world_size,
-    split_tensor_along_last_dim,
-    tensor_model_parallel_all_gather,
 )
 from sglang.srt.layers.communicator import LayerCommunicator, LayerScatterModes
 from sglang.srt.layers.dp_attention import get_attention_tp_rank, get_attention_tp_size