Update nightly evaluation models to include Qwen3-30B-A3B-Thinking-2507 and adjust test configurations accordingly

Author: HaiShaw

python/sglang/test/test_utils.py

@@ -118,7 +118,7 @@
 
 # Nightly tests
 DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP1 = "meta-llama/Llama-3.1-8B-Instruct,mistralai/Mistral-7B-Instruct-v0.3,deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct,google/gemma-2-27b-it"
-DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP2 = "meta-llama/Llama-3.1-70B-Instruct,mistralai/Mixtral-8x7B-Instruct-v0.1,Qwen/Qwen2-57B-A14B-Instruct"
+DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP2 = "meta-llama/Llama-3.1-70B-Instruct,mistralai/Mixtral-8x7B-Instruct-v0.1,Qwen/Qwen2-57B-A14B-Instruct,Qwen/Qwen3-30B-A3B-Thinking-2507"
 DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_FP8_TP1 = "neuralmagic/Meta-Llama-3.1-8B-Instruct-FP8,neuralmagic/Mistral-7B-Instruct-v0.3-FP8,neuralmagic/DeepSeek-Coder-V2-Lite-Instruct-FP8,neuralmagic/gemma-2-2b-it-FP8"
 DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_FP8_TP2 = "neuralmagic/Meta-Llama-3.1-70B-Instruct-FP8,neuralmagic/Mixtral-8x7B-Instruct-v0.1-FP8,neuralmagic/Qwen2-72B-Instruct-FP8,neuralmagic/Qwen2-57B-A14B-Instruct-FP8,neuralmagic/DeepSeek-Coder-V2-Lite-Instruct-FP8,zai-org/GLM-4.5-Air-FP8"
 DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_QUANT_TP1 = "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4,hugging-quants/Meta-Llama-3.1-8B-Instruct-GPTQ-INT4,hugging-quants/Mixtral-8x7B-Instruct-v0.1-AWQ-INT4"

test/srt/nightly/test_gsm8k_eval_amd.py

@@ -54,9 +54,8 @@ def remove_failing_models(model_str):
 DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP1 = remove_failing_models(
     DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP1
 )
-# AMD-specific models verified on MI300X
 DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP2 = remove_failing_models(
-    DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP2 + ",Qwen/Qwen3-30B-A3B-Thinking-2507"
+    DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_TP2
 )
 DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_FP8_TP1 = remove_failing_models(
     DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_FP8_TP1