feat: support loading eetq quantized model #393
Draft
Conversation
This file contains 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
|
will revisit again after the #399 resolved. |
|
@SidaZh is that possible help have a check this integration? |
SidaZh
reviewed
Apr 15, 2024
| @@ -226,6 +226,15 @@ def get_multi_weights_col(self, prefixes: List[Union[str, Tuple]], quantize: str | |||
|
|
|||
| bits, groupsize = self._get_gptq_params() | |||
| weight = (qweight, qzeros, scales, g_idx, bits, groupsize, False) | |||
| elif quantize == "eetq": | |||
| try: | |||
| qweight = torch.cat(self.get_sharded_list("qweight", prefixes, dim=1), dim=1) | |||
There was a problem hiding this comment.
Is it necessary to merge the weight parameters of multiple cards here? The quantization of eetq includes two steps: quantization and cutlass relayout. If the tensor is sliced or concat, the layout will be destroyed and output will be wrong.
There was a problem hiding this comment.
There are two options: 1. Save and load using the same tp and pp strategies; 2. Load basic per-channel quantization parameters and relayout when initializing EETQLinear which requires some development. The following two interfaces will help:
from EETQ import quant_weights, preprocess_weights
unprocessed_quantized_weight, processed_quantized_weight, scales = quant_weights(unquantized_weight, torch.int8, True) # quantize and relayout
processed_quantized_weight = preprocess_weights(unprocessed_quantized_weight) # relayout
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
What does this PR do?
Fixes #391
Before submitting
to it if that's the case.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@tgaddair thanks