Does TensorRT currently support quantization to bfloat16? I don’t see such an option. If not will there be that option in the future? Thanks!
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| TensoRT convewrsion in bfloat16 | 3 | 707 | February 1, 2024 | |
| ONNX TensorRT Engines FP16/32 | 5 | 202 | October 23, 2025 | |
| Quantization in TensorRt | 6 | 2152 | March 2, 2022 | |
| Why inference in jetson nano with fp16 is slower than fp32 | 9 | 2148 | September 5, 2021 | |
| Turing Tensor core int4 operation | 3 | 2957 | December 11, 2018 | |
| tensorRT FP8 support | 2 | 3113 | June 21, 2023 | |
| Unable to quantization FP8 in TensorRT | 1 | 675 | June 20, 2023 | |
| resnext in nano jetson | 10 | 1270 | October 14, 2021 | |
| TensorRT, result error in fp16 | 1 | 771 | October 19, 2021 | |
| Check device supports FP16 | 1 | 1488 | April 5, 2022 |