1 paper · 1 filter
Ashitabh Misra, Madhav Agrawal, Arham Jain +1
Mixed-precision quantization improves the accuracy of post-training quantization by allocating higher bitwidths to sensitive layers, but existing methods solve the allocation for a…