SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2307.10554 · ICCV · 2023

EMQ: Evolving Training-free Proxies for Automated Mixed Precision Quantization

Hengyue Pan, Xin Niu, Lujun Li, Peijie Dong, Zhiliang Tian, Zimian Wei

arXiv · PDF · Open in the Atlas

Code that ran

7 of 19 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, 2 violated it and 4 have no contract verdict; 1 failure of our instrument; 12 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/2307.10554.json · badge: /api/ran/2307.10554.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
lilujunai/emq-series marked official in Papers with Code
Papers with Code archive
7 ran · 12 no recorded run
FunctionStatusWhere it lives
BasicBlock Ran
ran on a synthesized input
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("cc18b58d20c93dfd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, 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
import torch.nn as nn
import numpy as np

# Define a simple BatchNorm stand-in for the BN referenced in the class
class BN(nn.Module):
    def __init__(self, num_features):
        super(BN, self).__init__()
        self.bn = nn.BatchNorm2d(num_features)
    
    def forward(self, x):
        return self.bn(x)

# Make BN available in the global scope for BasicBlock to use
globals()['BN'] = BN

# INIT_ARGS for BasicBlock.__init__
INIT_ARGS = {
    'inplanes': 16,
    'planes': 32,
    'stride': 1,
    'downsample': nn.Sequential(
        nn.Conv2d(16, 32, kernel_size=1, stride=1, bias=False),
        BN(32)
    ),
    'groups': 1,
    'base_width': 64,
    'dilation': 1,
    'norm_layer': BN
}

# ARGS for BasicBlock.forward
# Input tensor shape: (batch_size, inplanes, height, width)
ARGS = {
    'x': torch.randn(2, 16, 8, 8)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
Bottleneck Ran
ran on a synthesized input
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("df082992e7ba0798")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, 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
import torch.nn as nn
import numpy as np

# Define a simple BatchNorm stand-in for BN
class BN(nn.Module):
    def __init__(self, num_features):
        super(BN, self).__init__()
        self.bn = nn.BatchNorm2d(num_features)
    
    def forward(self, x):
        return self.bn(x)

# Create downsample module
downsample = nn.Sequential(
    nn.Conv2d(16, 64, kernel_size=1, stride=1, bias=False),
    BN(64)
)

INIT_ARGS = {
    'inplanes': 16,
    'planes': 16,
    'stride': 1,
    'downsample': downsample,
    'groups': 1,
    'base_width': 64,
    'dilation': 1,
    'norm_layer': BN
}

ARGS = {
    'x': torch.randn(2, 16, 8, 8)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
InvertedResidual Ran
ran on a synthesized input
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("491ef5b3386ab4b7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, 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
import torch.nn as nn

# INIT_ARGS: parameters for InvertedResidual.__init__
INIT_ARGS = {
    'inplanes': 16,
    'outplanes': 24,
    'stride': 1,
    'expand_ratio': 6
}

# ARGS: parameters for the forward method
ARGS = {
    'x': torch.randn(2, 16, 8, 8)  # batch=2, channels=16 (matches inplanes), height=8, width=8
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
StraightThrough Ran
ran on a synthesized input
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("78941fdc9f230be6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, dtype family float, shape (2, 3, 4, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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

class StraightThrough(nn.Module):
    """Return the input"""

    def __init__(self, channel_num: int=1):
        super().__init__()

    def forward(self, input):
        return input


INIT_ARGS = {
    "channel_num": 3
}

ARGS = {
    "input": torch.randn(2, 3, 4, 5)
}

if __name__ == "__main__":
    model = StraightThrough(**INIT_ARGS)
    output = model(**ARGS)
    print("Output shape:", output.shape)
    print("Output matches input:", torch.equal(output, ARGS["input"]))
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (2, 3, 4, 5) — shape (2, 3, 4, 5) as a later run recorded it
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
_fold_bn Ran
ran; our contract draft was wrong, not the code
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("140878966d2be929")
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. 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 Conv2d module
conv_module = nn.Conv2d(in_channels=3, out_channels=4, kernel_size=3, bias=True)

# Create a BatchNorm2d module
bn_module = nn.BatchNorm2d(num_features=4, affine=True)

# Initialize the modules with some data to populate running statistics
# We need to run forward pass to populate running_mean and running_var
x = torch.randn(2, 3, 8, 8)
conv_output = conv_module(x)
bn_module(conv_output)

ARGS = {
    "conv_module": conv_module,
    "bn_module": bn_module,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
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
is_absorbing Ran
ran, violated the contract we drafted
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("21b513cf47621143")
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 bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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)

# The function is_absorbing checks if m is an instance of Conv2d or Linear
# It takes a single parameter 'm' which should be a module

# Create a simple Conv2d module as a test case
conv_module = nn.Conv2d(in_channels=3, out_channels=4, kernel_size=3)

ARGS = {
    "m": conv_module
}
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 bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
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
is_bn Ran
ran, violated the contract we drafted
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("39d986a037856c1d")
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 bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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)

# The function is_bn checks if a module m is an instance of BatchNorm2d or BatchNorm1d
# It takes one parameter: m (a module)

# Create a BatchNorm2d module as a test case
bn_module = nn.BatchNorm2d(4)

ARGS = {
    "m": bn_module
}
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 bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
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
BaseQuantBlock Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("819d5ef809058894")
How it ran

no invocation recorded.

ComputeMadnasScore Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("2028d48de55aba38")
How it ran

no invocation recorded.

QuantBasicBlock Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("affd5495e8ec4d0e")
How it ran

no invocation recorded.

QuantBottleneck Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("88f5b8c6e343729d")
How it ran

no invocation recorded.

QuantInvertedResidual Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("47fc8538a2628e4c")
How it ran

no invocation recorded.

QuantModel Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("4895c833da6c7c80")
How it ran

no invocation recorded.

QuantModule Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("0d7bbdf53ee6e6db")
How it ran

no invocation recorded.

UniformAffineQuantizer Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("d4a79d5ee2cd4666")
How it ran

no invocation recorded.

compute_qe_score Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("ac0703e88e81917a")
How it ran

no invocation recorded.

convert_qnn Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("ced4515523449616")
How it ran

no invocation recorded.

fold_bn_into_conv Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("529664956b8dab9a")
How it ran

no invocation recorded.

search_fold_and_remove_bn Not yet run
unverified (harvested, no recorded run)
lilujunai/emq-series/emq/predictor/pruners/measures/qe_score.py
code served (permissive licence) · get_code("de6383d946b71e09")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 19 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 3 later) and 16 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
StraightThrough
brick of this paper's assembly: helper
4 other papers
1 dated earlier, 3 later · 3 only as an assembly brick
2302.04304 (2023-02)
Q-Diffusion: Quantizing Diffusion Models
2310.03270 (2023-10)
EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Models
2405.16005 (2024-05)
PTQ4DiT: Post-training Quantization for Diffusion Transformers
and 1 more not named here
BaseQuantBlock
brick of this paper's assembly: helper
3 other papers
1 dated earlier, 2 later
2302.04304 (2023-02)
Q-Diffusion: Quantizing Diffusion Models
2405.16005 (2024-05)
PTQ4DiT: Post-training Quantization for Diffusion Transformers
2506.22463 (2025-06)
Modulated Diffusion: Accelerating Generative Modeling with Modulated Quantization
search_fold_and_remove_bn
brick of this paper's assembly: helper
3 other papers
1 dated earlier, 1 later, 1 same year, order not determinable
Liu_PD-Quant_Post-Training_Quantization_Based_on_Prediction_Difference_Metric_CVPR_2023_paper (2023 (year only))
PD-Quant: Post-Training Quantization Based on Prediction Difference Metric
2303.11906 (2023-03)
Solving Oscillation Problem in Post-Training Quantization Through a Theoretical Perspective
2312.06372 (2023-12)
Ternary Spike: Learning Ternary Spikes for Spiking Neural Networks

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

Mixed-Precision Quantization (MQ) can achieve a competitive accuracy-complexity trade-off for models. Conventional training-based search methods require timeconsuming candidate training to search optimized per-layer bit-width configurations in MQ. Recently, some trainingfree approaches have presented various MQ proxies and significantly improve search efficiency. However, the correlation between these proxies and quantization accuracy is poorly understood. To address the gap, we first build the MQ-Bench-101, which involves different bit configurations and quantization results. Then, we observe that the existing training-free proxies perform weak correlations on the MQ-Bench-101. To efficiently seek superior proxies, we develop an automatic search of proxies framework for MQ via evolving algorithms. In particular, we devise an elaborate search space involving the existing proxies and perform an evolution search to discover the best correlated MQ proxy. We proposed a diversity-prompting selection strategy and compatibility screening protocol to avoid premature convergence and improve search efficiency. In this way, our Evolving proxies for Mixed-precision Quantization (EMQ) framework allows the auto-generation of proxies without heavy tuning and expert knowledge. Extensive experiments on ImageNet with various ResNet and MobileNet families demonstrate that our EMQ obtains superior performance than state-of-the-art mixed-precision methods at a significantly reduced cost. The code will be released.

For agents

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

get_harvested_code_for_paper("2307.10554")
get_code_for_paper("2307.10554")
have("2307.10554")

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

curl https://syntology.ai/api/ran/2307.10554.json

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

[![Syntology run record](https://syntology.ai/api/ran/2307.10554.svg)](https://syntology.ai/paper/2307.10554)

Connect an agent — have() is free.