True 4-Bit Quantized CNN Training on CPU – 92.34% on Cifar-10

True 4-Bit Quantized CNN Training on CPU – 92.34% on Cifar-10(arxiv.org)

3 points by shivnathtathe 62 days ago | 3 comments

> true 4-bit precision

This isn't one of the new block floating point schemes, it's bona fide 4-bit precision weights. It boggles my mind that can actually work.

yorwba 62 days ago | |

Well, the weights are accumulated in full precision and are multiplied by a full-precision scale factor after quantization, and the activations and backward pass are computed in full precision as well, so it's not quite true 4-bit precision training. The resulting model can be stored with just slightly more than 4 bits per parameter, though.

jcalvinowens 62 days ago | | |

I really just don't understand how the quantization error doesn't ruin the results. Is there some reading you'd recommend?

I can easily understand how the block formats win.