mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-02-21 10:15:46 +08:00
3c5aec19c2
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> |
||
|---|---|---|
| .. | ||
| _torch | ||
| api_stability | ||
| bindings | ||
| disaggregated | ||
| llmapi | ||
| others | ||
| scaffolding | ||
| tools | ||
| trt | ||
| utils | ||
| conftest.py | ||
| dump_checkpoint_stats.py | ||
| gc_utils.py | ||
| profile_utils.py | ||
| pytest.ini | ||
| test_model_runner_cpp.py | ||
| test_pip_install.py | ||