zhhuang-nv 97bc680cd8 feat: support kv cache reuse for MLA (#3571) ... * support kv cache reuse for MLA load compressed_kv and k_pe and do up-projection use 192/128 head size MLA context kernel support Blackwell and Hopper now Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * add CI test Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * fix: set k_pe head_num to 1 for kernel 2 and kernel 2V2 Signed-off-by: Mingyang Jiang <13463932+jmydurant@users.noreply.github.com> * resolve comments Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * use GPTJ style RoPE for MLA Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * fix rebase error and some docs Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * fix kv_lens Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * tiny fix Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * fix torch compile Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * fix: use normal device memory instead of pinned memory for unit test Signed-off-by: Mingyang Jiang <13463932+jmydurant@users.noreply.github.com> * fix L0 tests Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * fix torch compile after rebase Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * resolve comments Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> * resolve comments again Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> --------- Signed-off-by: Zhen Huang <145532724+zhhuang-nv@users.noreply.github.com> Signed-off-by: Mingyang Jiang <13463932+jmydurant@users.noreply.github.com> Signed-off-by: zhhuang-nv <145532724+zhhuang-nv@users.noreply.github.com> Co-authored-by: Mingyang Jiang <13463932+jmydurant@users.noreply.github.com>		2025-05-15 15:22:21 +08:00
..
attention_backend	feat: support kv cache reuse for MLA (#3571)	2025-05-15 15:22:21 +08:00
auto_deploy	Breaking change: perf: Enable scheduling overlap by default (#4174)	2025-05-15 14:27:36 +08:00
compilation	[TRTLLM-3105][feat] Add Piecewise CUDA Graph Support (#3804)	2025-05-09 11:04:01 +08:00
custom_ops	feat: support kv cache reuse for MLA (#3571)	2025-05-15 15:22:21 +08:00
distributed	Revert "feat: Low Precision Allreduce for PCIe based GPU" (#4340)	2025-05-15 09:52:39 +08:00
models	[TRTLLM-3330][feat] Support DeepSeek-R1 W4A8 on Hopper (#4123)	2025-05-14 15:48:07 +08:00
modules	feat: support kv cache reuse for MLA (#3571)	2025-05-15 15:22:21 +08:00
peft	feat: support multi lora adapters and TP (#3885)	2025-05-08 23:45:45 +08:00
pyexecutor	feat: support kv cache reuse for MLA (#3571)	2025-05-15 15:22:21 +08:00
speculative	[fix] Fix relaxed acceptance to support enabling it in context phase (#4126)	2025-05-09 14:11:14 +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 multi lora adapters and TP (#3885)	2025-05-08 23:45:45 +08:00
pipeline_interface.py	chore: bump version to 0.19.0 (#3598) (#3841)	2025-04-29 16:57:22 +08:00
utils.py	[TRTLLM-3105][feat] Add Piecewise CUDA Graph Support (#3804)	2025-05-09 11:04:01 +08:00