[AMD] Add MiniMax-M2.7 accuracy and performance nightly tests (#22722)

Co-authored-by: HaiShaw <hixiao@gmail.com>

python/sglang/srt/models/minimax_m2.py

@@ -33,7 +33,6 @@ from sglang.jit_kernel.all_reduce import (
 from sglang.kernel_api_logging import debug_kernel_api
 from sglang.srt.batch_overlap.two_batch_overlap import model_forward_maybe_tbo
 from sglang.srt.distributed import (
-    get_bool_env_var,
     get_moe_expert_parallel_world_size,
     get_pp_group,
     get_tensor_model_parallel_world_size,
@@ -81,9 +80,16 @@ from sglang.srt.model_loader.weight_utils import (
     maybe_remap_kv_scale_name,
 )
 from sglang.srt.server_args import get_global_server_args
+
+# get_bool_env_var is defined in sglang.srt.utils.common, not sglang.srt.distributed.
+# Importing from the wrong module causes this file to fail import, which prevents the
+# native MiniMaxM2ForCausalLM from registering in ModelRegistry. The fallback to the
+# transformers wrapper then crashes on config.rope_parameters (transformers v5 issue).
+# Other files (custom_all_reduce.py, hf_transformers_utils.py) also use sglang.srt.utils.
 from sglang.srt.utils import (
     BumpAllocator,
     add_prefix,
+    get_bool_env_var,
     get_compiler_backend,
     is_cuda,
     is_non_idle_and_non_empty,