kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-01-26 05:32:57 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
3c5aec19c2
TensorRT-LLMs/tensorrt_llm/_torch
History
Gal Hubara-Agam 3c5aec19c2
[#5048][enhance] AutoDeploy: Optimize prepare_inputs (#6634) 
Optimize prepare_inputs routine in AutoDeploy, as part of the effort to reduce the performance gap compared to the default backend.
This PR includes two major fixes, and some other minor tweaks:
1. Avoid back and forth data copies
2. Optimize position ids update by separating the implementation for generation mode and context mode.

Signed-off-by: Suyog Gupta <41447211+suyoggupta@users.noreply.github.com>
Signed-off-by: Gal Hubara Agam <96368689+galagam@users.noreply.github.com>
Co-authored-by: Suyog Gupta <41447211+suyoggupta@users.noreply.github.com>
2025-08-10 13:55:04 +03:00
..
attention_backend [None] [feat] Add model gpt-oss (#6645) 2025-08-07 03:04:18 -04:00
auto_deploy [#5048][enhance] AutoDeploy: Optimize prepare_inputs (#6634) 2025-08-10 13:55:04 +03:00
compilation [https://nvbugs/5252313][fix] Fix torch compile + MTP (#6554) 2025-08-05 10:31:29 -04:00
custom_ops [TRTLLM-6744][feat] Remove input_sf swizzle for module WideEPMoE (#6231) 2025-08-08 11:13:42 +08:00
debug Add debug hook to support dump tensor data and add new debug functions easily (#5182) 2025-06-24 17:45:28 +08:00
distributed [None][feat] Add NCCL Symmetric Integration for All Reduce (#4500) 2025-08-07 17:28:14 -07:00
models [Perf] Improve Llama4 performance for small max_seqlen cases (#6306) 2025-08-09 02:58:31 -04:00
modules [None][fix] WAR GPT OSS on H20 with Triton MOE (#6721) 2025-08-08 19:47:09 -04:00
peft feat: support multi lora adapters and TP (#3885) 2025-05-08 23:45:45 +08:00
pyexecutor [TRTLLM-6637][feat] Resolve KV cache divergence issue (#6628) 2025-08-09 23:15:04 +08:00
shared_tensor [1/N][TRTLLM-5195][feat] Share PyTorch tensor between processes (#5396) 2025-07-10 05:12:53 +09:00
speculative [None][feat] Optimize CUDA graph memory usage for spec decode cases (#6718) 2025-08-08 13:56:53 -04:00
__init__.py [nvbugs/5401156][fix] Avoid import all models when import trtllm._common (#6266) 2025-07-27 23:29:21 -04:00
autotuner.py chore: Improve the AutoTuner log information. (#6368) 2025-08-01 09:19:52 +08:00
expert_statistic.py Add MTP support for Online EPLB (#5213) 2025-06-25 07:58:13 +08:00
llm.py [TRTLLM-5208][BREAKING CHANGE] chore: make pytorch LLM the default (#5312) 2025-06-20 03:01:10 +08:00
metadata.py feat: no-cache attention in PyTorch workflow (#3085) 2025-04-05 01:54:32 +08:00
model_config.py [None][feat] Add NCCL Symmetric Integration for All Reduce (#4500) 2025-08-07 17:28:14 -07:00
utils.py [None] [feat] Add model gpt-oss (#6645) 2025-08-07 03:04:18 -04:00
virtual_memory.py [TRTLLM-4406][feat] LLM sleep & wakeup Part 1: virtual device memory (#5034) 2025-08-04 13:51:01 +08:00