Papers › QDrop: Randomly Dropping Quantization for Extremely Low-bit Post-Training Quantization

QDrop: Randomly Dropping Quantization for Extremely Low-bit Post-Training Quantization

11 Mar 2022arXiv:2203.05740archive 2025-07-28

Xiuying Wei, Ruihao Gong, Yuhang Li, Xianglong Liu, Fengwei Yu

Recently, post-training quantization (PTQ) has driven much attention to produce efficient neural networks without long-time retraining. Despite its low cost, current PTQ works tend to fail under the extremely low-bit setting. In this study, we pioneeringly confirm that properly incorporating activation quantization into the PTQ reconstruction benefits the final accuracy. To deeply understand the inherent reason, a theoretical framework is established, indicating that the flatness of the optimized low-bit model on calibration and test data is crucial. Based on the conclusion, a simple yet effective approach dubbed as QDROP is proposed, which randomly drops the quantization of activations during PTQ. Extensive experiments on various tasks including computer vision (image classification, object detection) and natural language processing (text classification and question answering) prove its superiority. With QDROP, the limit of PTQ is pushed to the 2-bit activation for the first time and the accuracy boost can be up to 51.49%. Without bells and whistles, QDROP establishes a new state of the art for PTQ. Our code is available at https://github.com/wimh966/QDrop and has been integrated into MQBench (https://github.com/ModelTC/MQBench)

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Syntology holds a reference implementation of QDROP, generated from the paper text (not the authors' code), at level 3 on Syntology's ladder: V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” (quoted from Syntology's server card). No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='QDROP') (how to connect); the free check have('2203.05740') needs no token.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 19 code samples harvested from 2 repositories linked to this paper: 14 ran (of which 5 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 1 violated, 5 with no contract checked; 8 where Syntology's instrument failed) · 5 unverified. Where Syntology's instrument failed: 8 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 19 samples from 2 repositories, 14 ran (6 with no instrument failure; 8 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed. 4 of the 14 samples that ran have no recorded date.

modeltc/mqbench officialmentioned in paperpytorchApache-2.0Syntology: 8 samples, 5 ran (4 instrument-failed) · 3 unverified report
wimh966/QDrop officialmentioned in paperpytorchSyntology: 11 samples, 9 ran (4 instrument-failed) · 2 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

19 samples harvested: 14 ran (of which 5 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 1 violated, 5 with no contract checked; 8 where Syntology's instrument failed) · 5 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · violated contract
8ran · our draft was wrong
5ran
5unverified

Licence: 0 of the 19 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. For 2 samples that ran, the graph records a signature this page does not show: for a class the fixture built without calling, the harness called the object with inputs it made up itself, so the recorded signature is the harness's probe, not the fixture's arguments. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

modeltc/mqbench official repository 8 samples, 5 ran (4 instrument-failed) · 3 unverified
layer_has_weights mqbench/advanced_ptq.py ran · violated contract Apache-2.0 (permissive) · 734e5f593576355a · report fixture: signature nodes:list/1/str, modules:dict/0/? · returned bool · level recorded, date not recorded
round_ste in 18 papers
census: 3 implementations from 6 papers, 1 distinct output on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
mqbench/fake_quantize/qdrop_quantizer.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 675da8641c602b03 · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded, date not recorded
_fake_quantize_learnable_per_tensor_affine_training mqbench/fake_quantize/qdrop_quantizer.py ran · our draft was wrong fingerprinted Apache-2.0 (permissive) · 7d7aa73f165ae52e · report fixture: signature x:Tensor/2/float, scale:Tensor/0/float, zero_point:Tensor/0/int, quant_min:Tensor/0/int, quant_max:Tensor/0/int · returned Tensor · level recorded 2026-09-22
node2modules mqbench/advanced_ptq.py ran · our draft was wrong Apache-2.0 (permissive) · ac274433bb73e20c · report fixture: signature name2modules:dict/0/?, nodes:list/1/MockNode · returned dict · level recorded, date not recorded
qnode2fpnode mqbench/advanced_ptq.py ran · our draft was wrong Apache-2.0 (permissive) · 2df773973fe13be2 · report fixture: signature quant_modules:list/1/MockNode, fp32_modules:list/1/MockNode · returned dict · level recorded, date not recorded
QDropFakeQuantize mqbench/fake_quantize/qdrop_quantizer.py unverified Apache-2.0 (permissive) · 21646d4b40b8c655 · report
QuantizeBase mqbench/fake_quantize/qdrop_quantizer.py unverified Apache-2.0 (permissive) · 7c23cc16e1d8d7b3 · report
_fake_quantize_learnable_per_channel_affine_training mqbench/fake_quantize/qdrop_quantizer.py unverified Apache-2.0 (permissive) · ff3b513126ce71ed · report
wimh966/QDrop official repository 11 samples, 9 ran (4 instrument-failed) · 2 unverified
DataSaverHook qdrop/solver/recon.py ran MIT (permissive) · aad5f7fa5a104f83 · report fixture: returned DataSaverHook · level recorded 2026-09-22
LSQPlusFakeQuantize qdrop/solver/recon.py ran fingerprinted MIT (permissive) · cae087e14b082ad7 · report fixture: returned LSQPlusFakeQuantize · level recorded 2026-09-22
LinearTempDecay qdrop/solver/recon.py ran MIT (permissive) · 4aded0dd4cff11f4 · report fixture: returned LinearTempDecay · level recorded 2026-09-22
LossFunction in 8 papers qdrop/solver/recon.py ran fingerprinted MIT (permissive) · 665e30ed4645891a · report fixture: returned LossFunction · level recorded 2026-09-22
QuantizedModule qdrop/solver/recon.py ran MIT (permissive) · f762b59f2c5ed75b · report fixture: returned QuantizedModule · level recorded 2026-09-22
fake_quantize_learnableplus_per_channel_affine_training qdrop/solver/recon.py ran · our draft was wrong fingerprinted MIT (permissive) · 7daaddceec308813 · report fixture: signature x:Tensor/4/float, scale:Tensor/1/float, zero_point:Tensor/1/float, ch_axis:int/0/int, quant_min:int/0/int, quant_max:int/0/int, grad_factor:float/0/float · returned Tensor · level recorded 2026-09-22
fake_quantize_learnableplus_per_tensor_affine_training qdrop/solver/recon.py ran · our draft was wrong fingerprinted MIT (permissive) · 128adf85beb85df4 · report fixture: signature x:Tensor/2/float, scale:Tensor/0/float, zero_point:Tensor/0/float, quant_min:int/0/int, quant_max:int/0/int, grad_factor:float/0/float · returned Tensor · level recorded 2026-09-22
lp_loss in 20 papers
census: 6 implementations from 10 papers, 1 distinct output on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32 · arg 2: rank 2, kind float, dtype float32
qdrop/solver/recon.py ran · our draft was wrong fingerprinted MIT (permissive) · 8003cb7a5176e497 · report fixture: signature pred:Tensor/2/float, tgt:Tensor/2/float, p:float/0/float · returned Tensor · level recorded 2026-09-22
save_inp_oup_data qdrop/solver/recon.py ran · our draft was wrong MIT (permissive) · 01e81fd5942011ad · report fixture: signature model:callable/0/?, module:callable/0/?, cali_data:Tensor/2/float, store_inp:bool/0/bool, store_oup:bool/0/bool, bs:int/0/int, keep_gpu:bool/0/bool · returned list · level recorded 2026-09-22
LSQFakeQuantize qdrop/solver/recon.py unverified MIT (permissive) · a1873dbf3ee330ef · report
reconstruction in 10 papers qdrop/solver/recon.py unverified MIT (permissive) · e62a3ec748b29b66 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. For a generated assembly, an independent peer is a second implementation generated separately from the same paper text; neither implementation read the authors' code. At level 2 the two agreed numerically on the fixture's inputs; level 3 is recorded when property tests derived from the paper also passed on both, and each level 3 card prints how many its report lists. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. A generated assembly's label was set when its code was generated for the method, and no model judged it; its card prints it as "by construction" or "not judged by a model". Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another. Where an assembly's phrase has no constructed figure, a brick that ran is not among the harvested samples this build reads, and whether a brick's run only constructed an object is read from that brick's harvested sample, so the figure was not counted; it does not say there were none.

3 assemblies on this paper: 2 harvested, 1 generated, listed in 2 groups (siblings lifted in one batch form one group).

generated Assembly for QDROP (graph method: QDROP), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer (level 3; 5 property tests passed on both).

label matches by construction: the code was generated for this method (not judged by a model)

  1. qdrop_backward_loss entry · level 3 · ran · honoured contract · 39403fc691a5f94bdb9c90e7b96098c54f08b6ca26b45c2d9fcf7c050ba05ff7

2 sibling assemblies for QDROP (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for QDROP (graph method: QDROP). 18 bricks; individually on a fixture: 16 ran (of which 8 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 8 where Syntology's instrument failed) · 2 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. reconstruction entry · level 0 · unverified · e62a3ec748b29b66
  2. LSQFakeQuantize helper · level 0 · unverified · a1873dbf3ee330ef
  3. LSQPlusFakeQuantize helper · level 1 · ran · cae087e14b082ad7
  4. LossFunction loss · level 1 · ran · 665e30ed4645891a
  5. QuantizeBase helper · level 1 · ran · f352d07e60e42108
  6. save_inp_oup_data helper · level 1 · ran · our draft was wrong · 01e81fd5942011ad
  7. fake_quantize_learnable_per_channel_affine_training helper · level 1 · ran · our draft was wrong · ba1ffd12796f5219
  8. fake_quantize_learnable_per_tensor_affine_training helper · level 1 · ran · our draft was wrong · 9eaea08ad40b7c03
  9. fake_quantize_learnableplus_per_channel_affine_training helper · level 1 · ran · our draft was wrong · 7daaddceec308813
  10. fake_quantize_learnableplus_per_tensor_affine_training helper · level 1 · ran · our draft was wrong · 128adf85beb85df4
  11. LinearTempDecay scheduler · level 1 · ran · 4aded0dd4cff11f4
  12. QuantizedModule helper · level 1 · ran · f762b59f2c5ed75b
  13. lp_loss metric · level 1 · ran · our draft was wrong · 8003cb7a5176e497
  14. ObserverBase helper · level 1 · ran · 119e117e97e53371
  15. DataSaverHook helper · level 1 · ran · aad5f7fa5a104f83
  16. StopForwardException helper · level 1 · ran · 13900bea8f3b93a1
  17. grad_scale helper · level 1 · ran · our draft was wrong · 043c3be8a95e065c
  18. round_ste helper · level 2 · ran · our draft was wrong · 675da8641c602b03

harvested Assembly for QDROP (graph method: QDROP). 5 bricks; individually on a fixture: 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 3 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. QDropFakeQuantize entry · level 0 · unverified · 21646d4b40b8c655
  2. QuantizeBase helper · level 0 · unverified · 7c23cc16e1d8d7b3
  3. _fake_quantize_learnable_per_channel_affine_training model · level 0 · unverified · ff3b513126ce71ed
  4. _fake_quantize_learnable_per_tensor_affine_training model · level 1 · ran · our draft was wrong · 7d7aa73f165ae52e
  5. round_ste helper · level 2 · ran · our draft was wrong · 675da8641c602b03

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2203.05740") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Image ClassificationObject DetectionQuantizationQuestion AnsweringText Classificationimage-classificationobject-detectiontext-classification

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections