SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.01721 · NeurIPS · 2024

DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs

Ying Wei, Linqi Song, Yichen Wu, Yingtao Zhang, Haokun Lin, Haobo Xu, Jingzhi Cui, Linzhan Mou, Zhenan Sun

arXiv · PDF · Open in the Atlas

Code that ran

10 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 10 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2406.01721.json · badge: /api/ran/2406.01721.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
Hsu1023/DuQuant canonical (classifier label)
classifier basis: census
10 ran · 3 no recorded run
FunctionStatusWhere it lives
build_model_and_tokenizer Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/generate_act_scale_shift.py
code served (permissive licence) · get_code("8c5df57a3d2b01ae")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from unittest.mock import Mock, MagicMock

torch.manual_seed(0)

# Create a mock tokenizer that satisfies AutoTokenizer.from_pretrained interface
mock_tokenizer = Mock()
mock_tokenizer.use_fast = False
mock_tokenizer.legacy = False

# Create a mock model that satisfies AutoModelForCausalLM.from_pretrained interface
mock_model = Mock()
mock_model.torch_dtype = torch.float16
mock_model.device_map = 'auto'

# Patch the AutoTokenizer and AutoModelForCausalLM to return our mocks
from transformers import AutoModelForCausalLM, AutoTokenizer

original_tokenizer_from_pretrained = AutoTokenizer.from_pretrained
original_model_from_pretrained = AutoModelForCausalLM.from_pretrained

def mock_tokenizer_from_pretrained(model_name, **kwargs):
    return mock_tokenizer

def mock_model_from_pretrained(model_name, **kwargs):
    return mock_model

AutoTokenizer.from_pretrained = mock_tokenizer_from_pretrained
AutoModelForCausalLM.from_pretrained = mock_model_from_pretrained

ARGS = {
    'model_name': 'gpt2'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_layers Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/models/models_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from utkarshsaxena1/eigenattn, is pointer only (licence: NONE) · get_code("904cd61df2fe8fb9")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from utkarshsaxena1/eigenattn, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

format_example Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/mmlu_eval.py
code served (permissive licence) · get_code("cd763eaf1ac287e7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd

# Create a small DataFrame with the structure expected by format_example
# The function expects:
# - Column 0: prompt text
# - Columns 1 to k: answer choices (k columns)
# - Column k+1: the correct answer
# With choices = ['A', 'B', 'C', 'D'], we have k=4 choices

data = {
    0: ["What is 2+2?"],
    1: ["3"],
    2: ["4"],
    3: ["5"],
    4: ["6"],
    5: ["B"]  # The answer is 'B' (which corresponds to choice index 1, value "4")
}

df = pd.DataFrame(data)

ARGS = {
    'df': df,
    'idx': 0,
    'include_answer': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
format_subject Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/mmlu_eval.py
code served (permissive licence) · get_code("6ab745408cb8648b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

def format_subject(subject):
    l = subject.split('_')
    s = ''
    for entry in l:
        s += ' ' + entry
    return s

ARGS = {
    'subject': 'machine_learning_model'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
gen_prompt Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/mmlu_eval.py
code served (permissive licence) · get_code("3383488ea82b6c15")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd
import numpy as np

# Create a small DataFrame that mimics the structure expected by gen_prompt
# The function expects: question in column 0, answer choices in columns 1 to k, answer in column k+1
data = {
    0: ['What is 2+2?', 'What is the capital of France?'],
    1: ['3', 'London'],
    2: ['4', 'Paris'],
    3: ['5', 'Berlin'],
    4: ['6', 'Madrid'],
    5: ['A', 'B']  # answers column
}

train_df = pd.DataFrame(data)

subject = 'basic_math'

ARGS = {
    'train_df': train_df,
    'subject': subject,
    'k': -1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_act_scales Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/generate_act_scale_shift.py
code served (permissive licence) · get_code("bd1c7aaaa05aeaf3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a simple model with Linear layers
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

# Create a simple dataloader as a list of (x, y) tuples
# Each input tensor has shape (2, 4) - batch_size=2, input_dim=4
dataloader = [
    (torch.randn(2, 4), torch.randint(0, 2, (2,)))
    for _ in range(128)
]

num_samples = 128

ARGS = {
    "model": model,
    "dataloader": dataloader,
    "num_samples": num_samples
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_act_shifts Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/generate_act_scale_shift.py
code served (permissive licence) · get_code("550d556e74fb20f7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a small model with Linear layers
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

# Create a small dataloader with (x, y) tuples
# Each sample should have batch size and feature dimension matching model input
dataloader = [
    (torch.randn(2, 4), torch.randn(2, 2)) for _ in range(128)
]

ARGS = {
    "model": model,
    "dataloader": dataloader,
    "num_samples": 128
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
make_disjoint_window Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/models/models_utils.py
code served (permissive licence) · get_code("4df9168e0a0c38eb")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# The function takes a pair (a, b) where both a and b are sequences
# It slices a to remove the last (len(b) - 1) elements
# Both a and b should support len() and slicing operations
# Lists work perfectly for this

a = [1, 2, 3, 4, 5, 6, 7, 8]  # Context window
b = [9, 10, 11]  # Continuation window

pair = (a, b)

ARGS = {
    "pair": pair
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
truncate_number Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/models/transformation.py
pointer only (licence: MIT); the stored copy get_code serves, from saic-fi/mobilequant, is pointer only (licence: NOASSERTION) · get_code("94587954e392b665")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from saic-fi/mobilequant, and GitHub could not classify the licence of that repository (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

truncate_number Ran
ran; our contract draft was wrong, not the code
Hsu1023/DuQuant/models/models_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from utkarshsaxena1/eigenattn, is pointer only (licence: NONE) · get_code("4e53fbbdfa160f7d")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from utkarshsaxena1/eigenattn, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_pile Not yet run
unverified (harvested, no recorded run)
Hsu1023/DuQuant/datautils.py
pointer only (licence: MIT); the stored copy get_code serves, from utkarshsaxena1/eigenattn, is pointer only (licence: NONE) · get_code("3b51341ed1a0b73e")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from utkarshsaxena1/eigenattn, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

get_ptb Not yet run
unverified (harvested, no recorded run)
Hsu1023/DuQuant/datautils.py
code served (permissive licence) · get_code("bd4f416e271c2fb5")
How it ran

no invocation recorded.

get_wikitext2 Not yet run
unverified (harvested, no recorded run)
Hsu1023/DuQuant/datautils.py
code served (permissive licence) · get_code("406294899f965c2f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 12 are also present in other papers whose code we hold (12 with at least one paper dated earlier, 2 the same month, 12 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
format_subject 21 other papers
12 dated earlier, 1 the same month, 8 later
2009.03300 (2020-09)
Measuring Massive Multitask Language Understanding
2304.12986 (2023-04)
Measuring Massive Multitask Chinese Understanding
2305.17306 (2023-05)
Chain-of-Thought Hub: A Continuous Effort to Measure Large Language Models' Reasoning Performance
and 18 more not named here
format_example 16 other papers
7 dated earlier, 2 the same month, 7 later
2009.03300 (2020-09)
Measuring Massive Multitask Language Understanding
2305.17306 (2023-05)
Chain-of-Thought Hub: A Continuous Effort to Measure Large Language Models' Reasoning Performance
2310.05736 (2023-10)
LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Models
and 13 more not named here
gen_prompt 10 other papers
6 dated earlier, 4 later
2009.03300 (2020-09)
Measuring Massive Multitask Language Understanding
2305.17306 (2023-05)
Chain-of-Thought Hub: A Continuous Effort to Measure Large Language Models' Reasoning Performance
2310.05736 (2023-10)
LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Models
and 7 more not named here
make_disjoint_window 10 other papers
7 dated earlier, 3 later
2304.01089 (2023-04)
RPTQ: Reorder-based Post-training Quantization for Large Language Models
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2310.08041 (2023-10)
QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Models
and 7 more not named here
find_layers 9 other papers
7 dated earlier, 2 later
2304.01089 (2023-04)
RPTQ: Reorder-based Post-training Quantization for Large Language Models
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2310.08041 (2023-10)
QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Models
and 6 more not named here
truncate_number 8 other papers
6 dated earlier, 2 later
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2310.08041 (2023-10)
QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Models
2402.11295 (2024-02)
OneBit: Towards Extremely Low-bit Large Language Models
and 5 more not named here
truncate_number 8 other papers
5 dated earlier, 3 later
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2310.08041 (2023-10)
QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Models
2403.12544 (2024-03)
AffineQuant: Affine Transformation Quantization for Large Language Models
and 5 more not named here
get_pile 7 other papers
5 dated earlier, 2 later
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2310.08041 (2023-10)
QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Models
2402.11295 (2024-02)
OneBit: Towards Extremely Low-bit Large Language Models
and 4 more not named here
get_ptb 6 other papers
5 dated earlier, 1 later
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2402.11295 (2024-02)
OneBit: Towards Extremely Low-bit Large Language Models
2403.12544 (2024-03)
AffineQuant: Affine Transformation Quantization for Large Language Models
and 3 more not named here
get_wikitext2 5 other papers
4 dated earlier, 1 later
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2402.11295 (2024-02)
OneBit: Towards Extremely Low-bit Large Language Models
2403.12544 (2024-03)
AffineQuant: Affine Transformation Quantization for Large Language Models
and 2 more not named here
get_act_scales 3 other papers
2 dated earlier, 1 later
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2403.12544 (2024-03)
AffineQuant: Affine Transformation Quantization for Large Language Models
2506.03781 (2025-06)
Unifying Uniform and Binary-coding Quantization for Accurate Compression of Large Language Models
get_act_shifts 3 other papers
2 dated earlier, 1 later
2308.13137 (2023-08)
OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Models
2403.12544 (2024-03)
AffineQuant: Affine Transformation Quantization for Large Language Models
2506.03781 (2025-06)
Unifying Uniform and Binary-coding Quantization for Accurate Compression of Large Language Models

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Quantization of large language models (LLMs) faces significant challenges, particularly due to the presence of outlier activations that impede efficient low-bit representation. Traditional approaches predominantly address Normal Outliers, which are activations across all tokens with relatively large magnitudes. However, these methods struggle with smoothing Massive Outliers that display significantly larger values, which leads to significant performance degradation in low-bit quantization. In this paper, we introduce DuQuant, a novel approach that utilizes rotation and permutation transformations to more effectively mitigate both massive and normal outliers. First, DuQuant starts by constructing the rotation matrix, using specific outlier dimensions as prior knowledge, to redistribute outliers to adjacent channels by block-wise rotation. Second, We further employ a zigzag permutation to balance the distribution of outliers across blocks, thereby reducing block-wise variance. A subsequent rotation further smooths the activation landscape, enhancing model performance. DuQuant simplifies the quantization process and excels in managing outliers, outperforming the state-of-the-art baselines across various sizes and types of LLMs on multiple tasks, even with 4-bit weight-activation quantization. Our code is available at https://github.com/Hsu1023/DuQuant.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2406.01721")
get_code_for_paper("2406.01721")
have("2406.01721")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2406.01721.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2406.01721.svg)](https://syntology.ai/paper/2406.01721)

Connect an agent — have() is free.