Merge branch 'main' into gk/improved_sharding_heuristics

Author: greg-kwasniewski1

tensorrt_llm/llmapi/llm_args.py

@@ -2719,7 +2719,7 @@ class TorchLlmArgs(BaseLlmArgs):
     _quant_config: Optional[QuantConfig] = PrivateAttr(default=None)
 
     disable_flashinfer_sampling: bool = Field(
-        default=True,
+        default=False,
         description=
         "Disable the use of FlashInfer.sampling. This option is likely to be removed in the future.",
         status="prototype",

tests/unittest/api_stability/references/llm.yaml

@@ -109,7 +109,7 @@ methods:
         status: beta
       disable_flashinfer_sampling:
         annotation: bool
-        default: True
+        default: False
         status: prototype
       moe_config:
         annotation: tensorrt_llm.llmapi.llm_args.MoeConfig