-
Notifications
You must be signed in to change notification settings - Fork 2.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
RNN-T and TDT inference: use CUDA graphs by default #8972
Commits on Apr 18, 2024
-
Use Cuda graphs by default for transcription
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 7441a2f - Browse repository at this point
Copy the full SHA 7441a2fView commit details -
RNN-T Loop Labels + Cuda graphs user-friendly
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 4761e68 - Browse repository at this point
Copy the full SHA 4761e68View commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 2b13f31 - Browse repository at this point
Copy the full SHA 2b13f31View commit details -
Configuration menu - View commit details
-
Copy full SHA for d391e98 - Browse repository at this point
Copy the full SHA d391e98View commit details -
Enable by default Cuda graphs for TDT
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 0205101 - Browse repository at this point
Copy the full SHA 0205101View commit details
Commits on Apr 19, 2024
-
Configuration menu - View commit details
-
Copy full SHA for 0cac599 - Browse repository at this point
Copy the full SHA 0cac599View commit details -
Configuration menu - View commit details
-
Copy full SHA for cb68701 - Browse repository at this point
Copy the full SHA cb68701View commit details
Commits on Apr 25, 2024
-
Configuration menu - View commit details
-
Copy full SHA for b38ff6f - Browse repository at this point
Copy the full SHA b38ff6fView commit details -
Configuration menu - View commit details
-
Copy full SHA for 25235bb - Browse repository at this point
Copy the full SHA 25235bbView commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for f175c8a - Browse repository at this point
Copy the full SHA f175c8aView commit details -
Configuration menu - View commit details
-
Copy full SHA for 1672739 - Browse repository at this point
Copy the full SHA 1672739View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9e18150 - Browse repository at this point
Copy the full SHA 9e18150View commit details -
Configuration menu - View commit details
-
Copy full SHA for 94ba6af - Browse repository at this point
Copy the full SHA 94ba6afView commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 7b9d619 - Browse repository at this point
Copy the full SHA 7b9d619View commit details -
Configuration menu - View commit details
-
Copy full SHA for 2d3b083 - Browse repository at this point
Copy the full SHA 2d3b083View commit details -
Configuration menu - View commit details
-
Copy full SHA for 7e805bc - Browse repository at this point
Copy the full SHA 7e805bcView commit details
Commits on Apr 27, 2024
-
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 43c01ac - Browse repository at this point
Copy the full SHA 43c01acView commit details -
Set max_symbols to 10 if None. Add comments
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 030be86 - Browse repository at this point
Copy the full SHA 030be86View commit details -
Fix issue with confidence + bfloat16
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 100fd9c - Browse repository at this point
Copy the full SHA 100fd9cView commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 6b5d1d2 - Browse repository at this point
Copy the full SHA 6b5d1d2View commit details -
Add comment about setting variables in config
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 07bd665 - Browse repository at this point
Copy the full SHA 07bd665View commit details -
Configuration menu - View commit details
-
Copy full SHA for 638823e - Browse repository at this point
Copy the full SHA 638823eView commit details
Commits on Apr 29, 2024
-
Enable CUDA graphs everywhere. Disable explicitly in training pipeline.
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 464dd51 - Browse repository at this point
Copy the full SHA 464dd51View commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 2b7cd73 - Browse repository at this point
Copy the full SHA 2b7cd73View commit details -
Configuration menu - View commit details
-
Copy full SHA for e730f91 - Browse repository at this point
Copy the full SHA e730f91View commit details -
Configuration menu - View commit details
-
Copy full SHA for cc06bf0 - Browse repository at this point
Copy the full SHA cc06bf0View commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for cf38241 - Browse repository at this point
Copy the full SHA cf38241View commit details -
Instantiate RNNTGreedyDecodeCudaGraph only when all conditions are met
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 19ca09d - Browse repository at this point
Copy the full SHA 19ca09dView commit details
Commits on Apr 30, 2024
-
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for d98b8fc - Browse repository at this point
Copy the full SHA d98b8fcView commit details -
Configuration menu - View commit details
-
Copy full SHA for 05eb103 - Browse repository at this point
Copy the full SHA 05eb103View commit details -
Move toggling CUDA graphs to
ASRModel
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 7c6f7f0 - Browse repository at this point
Copy the full SHA 7c6f7f0View commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for cb6d500 - Browse repository at this point
Copy the full SHA cb6d500View commit details -
Configuration menu - View commit details
-
Copy full SHA for 35564df - Browse repository at this point
Copy the full SHA 35564dfView commit details -
Configuration menu - View commit details
-
Copy full SHA for 7192acc - Browse repository at this point
Copy the full SHA 7192accView commit details -
Configuration menu - View commit details
-
Copy full SHA for d4a27f6 - Browse repository at this point
Copy the full SHA d4a27f6View commit details
Commits on May 2, 2024
-
Extract toggling CUDA graphs logic to
WithOptionalCudaGraphs
. Fix C……UDA graphs in `ASRModel` Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for c0877f2 - Browse repository at this point
Copy the full SHA c0877f2View commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for b880510 - Browse repository at this point
Copy the full SHA b880510View commit details -
Configuration menu - View commit details
-
Copy full SHA for 82f83dc - Browse repository at this point
Copy the full SHA 82f83dcView commit details -
Signed-off-by: Vladimir Bataev <vbataev@nvidia.com>
Configuration menu - View commit details
-
Copy full SHA for 4e47010 - Browse repository at this point
Copy the full SHA 4e47010View commit details