In tensorRT quantization, the candidate distribution Q which has 128bins is afterwards expanded into i bins , Here, How does it work in this process? if candidate_Q = [10 ,8 , 8 , 7 ,5] and into 7 bins , how can the 7bins generate from candidate_Q?
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
tensorRT quantization | 0 | 652 | January 31, 2018 | |
How to form the Q-distribution in tensorrt 8bit Calibration? | 0 | 911 | December 24, 2018 | |
Three questions about 8-bit quantization(Entropy Calibration - pseudocode) | 0 | 1801 | December 20, 2018 | |
Quantization to int8 still confusing | 1 | 491 | March 23, 2020 | |
Tensorrt - Entropy Calibration - pseudocode | 5 | 2596 | January 9, 2020 | |
Questions on TensorRT 2.1.2 | 2 | 666 | January 17, 2018 | |
Questions for PTQ Entropy Calibration | 1 | 687 | November 22, 2021 | |
Examples of more diverse usage of TensorRT IInt8Calibrator available elsewhere? | 0 | 493 | July 21, 2017 | |
TensorRT help for beginners | 1 | 1027 | August 30, 2018 | |
Int8 quantization | 1 | 499 | December 16, 2021 |