Asymmetric quantized matmul support #7626
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode characters
update test update readme fix test add ssymmetric quant op support
lsy323 force-pushed the lsiyuan/asymmetric-quant branch from c5e9081 to 69dc1e8 Compare 
JackCaoG approved these changes Jul 9, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit. This suggestion is invalid because no changes were made to the code. Suggestions cannot be applied while the pull request is closed. Suggestions cannot be applied while viewing a subset of changes. Only one suggestion per line can be applied in a batch. Add this suggestion to a batch that can be applied as a single commit. Applying suggestions on deleted lines is not supported. You must change the existing code in this line in order to create a valid suggestion. Outdated suggestions cannot be applied. This suggestion has been applied or marked resolved. Suggestions cannot be applied from pending reviews. Suggestions cannot be applied on multi-line comments. Suggestions cannot be applied while the pull request is queued to merge. Suggestion cannot be applied right now. Please check back later.
This PR depends on #7605 to land first
With asymmetric quantization,
w_dq = w_int * weight_scaler - zero_point.Thus the matmul becomes
mamtul_out = x @ w_int * weight_scaler - x @ zero_point.unsqueeze(0).broadcast(x.shape[-1])To compute the item
x @ zero_point.unsqueeze(0).broadcast(x.shape[-1]), we useeinsum('...c, z', x, zero_point)for per-channel quant, andmatmul(x.sum(-1), zero_point)for blockwise quant.