Fix broken HQQ support#45147
Open
mobicham wants to merge 7 commits intohuggingface:mainfrom
Open
Conversation
Contributor
Author
|
@ArthurZucker @SunMarc a little bump on this, should be an easy fix |
Contributor
|
[For maintainers] Suggested jobs to run (before merge) run-slow: hqq |
SunMarc
reviewed
Apr 9, 2026
Member
SunMarc
left a comment
SunMarc
left a comment
There was a problem hiding this comment.
Thanks for looking into that @mobicham ! Since we are adding back the support, can we try to do something a bit more maintainable ? This is also one of the reason we didn't add back the support, it was too complicated. I know that SINQ based a lot of their code on hqq / gemlite and the first PR they added did something similar to here but in the end they manage to clean a lot the integration and now it looks much better. Would you be up to do that ? https://github.com/huggingface/transformers/blob/main/src/transformers/quantizers/quantizer_sinq.py
Comment on lines
165
to
168
| # TODO: to remove | ||
| # def create_quantized_param( | ||
| # self, | ||
| # model: "PreTrainedModel", |
Comment on lines
+162
to
+163
| def get_weight_conversions(self): | ||
| return [] |
Comment on lines
91
to
94
| # TODO: to remove | ||
| # Kept here in case we see some interest in adding support for it | ||
| # # Adds missing keys for HQQLinear modules that are loaded but the model with initialized with torch.nn.Linear | ||
| # def update_expected_keys( |
Comment on lines
73
to
78
| @@ -72,6 +78,7 @@ | |||
| logger.info("Setting dtype to torch.float32 as the default value since it was not specified.") | |||
Comment on lines
+67
to
+70
| def update_dtype(self, dtype): | ||
| if dtype is not None: | ||
| self.dtype = dtype | ||
| return dtype |
Member
There was a problem hiding this comment.
do we really need that, the tensors should be in the right dtype so we should be able to access that directly
Comment on lines
+335
to
+347
| def _load_hqq_from_checkpoint(self, model: "PreTrainedModel"): | ||
| """Load pre-quantized HQQ weights directly from checkpoint files.""" | ||
| from collections import defaultdict | ||
|
|
||
| from safetensors import safe_open | ||
|
|
||
| from ..integrations.hqq import autoname_modules, name_to_linear_tag | ||
|
|
||
| # Determine target device from stored device_map | ||
| device_map = getattr(self, "device_map", None) | ||
| if isinstance(device_map, dict): | ||
| # Use the first non-cpu device from the map (values can be str, int, or torch.device) | ||
| devices = [torch.device(v) for v in device_map.values()] |
Comment on lines
+248
to
+261
| def _setup_missing_key_filters(self, model, checkpoint_files): | ||
| """Scan checkpoint files to find HQQ-quantized modules. | ||
|
|
||
| For those modules: | ||
| 1. Suppress their .weight missing key warnings in the load report. | ||
| 2. Replace their weight parameter with a scalar meta tensor so that | ||
| ``_move_missing_keys_from_meta_to_device`` does not allocate | ||
| full-size fp16 tensors on GPU (which would cause OOM). | ||
| """ | ||
| import re | ||
|
|
||
| from safetensors import safe_open | ||
|
|
||
| quantized_modules = set() |
Member
There was a problem hiding this comment.
Can we not do these types of changes ?
Comment on lines
+161
to
+182
| quant_config = getattr(module, "quant_config", None) | ||
| if quant_config is None: | ||
| # Module is skipped from quantization, just return the weight as-is | ||
| return {full_layer_name: value} | ||
|
|
||
| # Determine target device and compute dtype | ||
| target_device = value.device | ||
| compute_dtype = self.hf_quantizer.dtype | ||
|
|
||
| # Create HQQLinear from the nn.Linear | ||
| hqq_layer = HQQLinear( | ||
| module, | ||
| quant_config=quant_config, | ||
| compute_dtype=compute_dtype, | ||
| device=target_device, | ||
| del_orig=True, | ||
| ) | ||
|
|
||
| if hqq_layer.bias is not None and isinstance(hqq_layer.bias, torch.Tensor): | ||
| hqq_layer.bias = torch.nn.Parameter(hqq_layer.bias) | ||
|
|
||
| if self.hf_quantizer.using_multi_gpu: |
Member
There was a problem hiding this comment.
It would be nice if we didn't have to do that here
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
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.
What does this PR do?
This PR fixes hqq support that has been broken for a couple of months now after a refactoring:
Code Agent Policy
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
@ArthurZucker @SunMarc