Looking for a way to accelerate inference for Tensorflow Saved Models. Is tehre a way to use AMP with TF Seaved Models ? FP16 for starters will be a significant progress.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| AMP didn't convert any node to float16 | 1 | 806 | July 19, 2019 | |
| TF-AMP being used with TensorRT? | 2 | 324 | June 10, 2024 | |
| How to enable AMP | 2 | 781 | July 14, 2021 | |
| Training Precision ISSUE | 2 | 514 | October 26, 2020 | |
| Where do i set enable_amp? | 2 | 510 | August 19, 2020 | |
| automatic mixed precision failure | 5 | 3285 | October 4, 2021 | |
| Automatic Mixed Precision for NVIDIA Tensor Core Architecture in TensorFlow | 5 | 639 | October 8, 2021 | |
| AMP error from tensorflow | 3 | 950 | April 11, 2019 | |
| How can i Using (Mixed Precision/ Automatic Mixed Precision) INT8(IMMA) | 0 | 634 | October 4, 2019 | |
| NVIDIA's apex only for inferences | 0 | 738 | March 19, 2020 |