Yukun He 83b36ebecd Fix fused_moe fallback issue. (#3652) ... min_latency_mode is only set to False during warmup phase. Thus when it becomes true during inference, all tactics fall back to the default one and thus cause perf regression. Signed-off-by: Yukun He <23156053+hyukn@users.noreply.github.com>		2025-04-17 23:17:04 +08:00
..
attention_backend	Support CUDA graphs for EAGLE3 (#3176)	2025-04-17 04:53:50 +08:00
auto_deploy	chore: move all distributed related codes into _torch.distributed directory (#3511)	2025-04-15 08:39:17 +08:00
compilation	feat: Add NVFP4 UB pattern optimization pass in torch compile (#3371)	2025-04-11 21:25:29 +08:00
custom_ops	Fix fused_moe fallback issue. (#3652)	2025-04-17 23:17:04 +08:00
distributed	chore: move all distributed related codes into _torch.distributed directory (#3511)	2025-04-15 08:39:17 +08:00
models	chore: Use ellipsis as default value to detect whether residual argument is provided (#3626)	2025-04-17 12:31:58 +08:00
modules	feat: Support unfused rope in MLA. (#3610)	2025-04-17 16:50:49 +08:00
peft	added loraOp into lora layer + test for mlp and comparison to lora plugin (#3455)	2025-04-17 12:48:27 +08:00
pyexecutor	feat: allocate minimal blocks per window size (#3028)	2025-04-17 16:04:57 +08:00
speculative	Support CUDA graphs for EAGLE3 (#3176)	2025-04-17 04:53:50 +08:00
__init__.py	Update TensorRT-LLM (#2755)	2025-02-11 03:01:00 +00:00
autotuner.py	feat: Apply the new torch-flow compatible AutoTuner to both Fused MoE and NVFP4 Linear operators. (#3151)	2025-04-08 14:28:36 +08:00
llm.py	test: [TRTLLM-4334] Create 1.0 criteria scope from API stability references (#3069)	2025-03-26 18:14:35 +08:00
metadata.py	feat: no-cache attention in PyTorch workflow (#3085)	2025-04-05 01:54:32 +08:00
model_config.py	feat: Support cos_sin_cache in all cases. (#3517)	2025-04-16 13:48:44 +08:00
pipeline_interface.py	Update (#2978)	2025-03-23 16:39:35 +08:00
utils.py	Cache sin cos in model instead of global LRU cache. (#3378)	2025-04-14 11:19:09 +08:00