Simplify the nan detection and greedy check in sampler (#1709)

2024-10-18 20:21:24 -07:00
parent 2bcfba1b08
commit f0f8a7699b
6 changed files with 24 additions and 7 deletions
--- a/python/sglang/srt/managers/schedule_batch.py
+++ b/python/sglang/srt/managers/schedule_batch.py
@@ -53,6 +53,7 @@ global_server_args_dict = {
    "triton_attention_reduce_in_fp32": ServerArgs.triton_attention_reduce_in_fp32,
    "disable_mla": ServerArgs.disable_mla,
    "torchao_config": ServerArgs.torchao_config,
+    "disable_nan_detection": ServerArgs.disable_nan_detection,
 }