SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2002.05712 · 2020

Cross-Iteration Batch Normalization

arXiv · PDF · Open in the Atlas

Code that ran

4 of 6 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 4 have no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-21; 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/2002.05712.json · badge: /api/ran/2002.05712.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
hlld/cross-iteration-batch_normalization community (classifier label)
classifier basis not recorded on the link
4 ran · 2 no recorded run
FunctionStatusWhere it lives
filter_regularization_variables Ran
ran on a synthesized input
hlld/cross-iteration-batch_normalization/utils.py
code served (permissive licence) · get_code("22863669fed87b1a")
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 list, 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 re
import torch
import torch.nn as nn

# Create a simple mock class to represent TensorFlow Variable
class MockVariable:
    def __init__(self, name):
        self.name = name

# Create test variables with different naming patterns
variables = [
    MockVariable('layer1/kernel:0'),
    MockVariable('layer1/bias:0'),
    MockVariable('layer2/weight:0'),
    MockVariable('layer2/bias:0'),
    MockVariable('layer3/kernel:0'),
    MockVariable('layer4/weight:0'),
    MockVariable('layer5/other:0'),
]

ARGS = {
    'variables': variables,
    'regularization_variable_regex': '.*(kernel|weight):0$',
}
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 list, ndim 1, dtype family float — a list 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
filter_trainable_variables Ran
ran on a synthesized input
hlld/cross-iteration-batch_normalization/utils.py
code served (permissive licence) · get_code("58c670dab5b13114")
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 list, 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 re

class MockVariable:
    """Mock TensorFlow variable with a name attribute."""
    def __init__(self, name):
        self.name = name

# Create a list of mock variables with different names
variables = [
    MockVariable("layer1/weights:0"),
    MockVariable("layer1/biases:0"),
    MockVariable("layer2/weights:0"),
    MockVariable("frozen_layer/weights:0"),
    MockVariable("frozen_layer/biases:0"),
    MockVariable("encoder/dense:0"),
]

# Define frozen_variable_prefix as a regex pattern
frozen_variable_prefix = "frozen_.*"

ARGS = {
    "variables": variables,
    "frozen_variable_prefix": frozen_variable_prefix,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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
filter_variables Ran
ran on a synthesized input
hlld/cross-iteration-batch_normalization/utils.py
code served (permissive licence) · get_code("2d51886963260f5e")
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 list, 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 re
import torch


class MockVariable:
    """Mock TensorFlow variable with a name attribute."""
    def __init__(self, name):
        self.name = name


# Create a list of mock variables with distinct names
variables = [
    MockVariable("layer1/weights:0"),
    MockVariable("layer1/bias:0"),
    MockVariable("layer2/weights:0"),
    MockVariable("layer2/bias:0"),
    MockVariable("optimizer/momentum:0"),
]

# Define test arguments
ARGS = {
    "variables": variables,
    "variable_regex": r"layer\d+/weights:0",
    "is_whitelist": True,
}
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 list, ndim 1, dtype family float — a list 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
load_CIFAR_batch Ran
ran on a synthesized input
hlld/cross-iteration-batch_normalization/cifar10.py
code served (permissive licence) · get_code("0387f93df835ab04")
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 tempfile
import os
import numpy as np
from six.moves import cPickle as pickle

# Create a temporary CIFAR-like batch file
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, 'cifar_batch')

# Create a minimal CIFAR batch dictionary
cifar_data = {
    'data': np.random.randint(0, 256, size=(10000, 3072), dtype=np.uint8),
    'labels': np.random.randint(0, 10, size=(10000,), dtype=np.int32).tolist()
}

# Pickle it to the temporary file
with open(temp_file, 'wb') as f:
    pickle.dump(cifar_data, f)

ARGS = {
    'filename': temp_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, six — 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
load_CIFAR10 Not yet run
unverified (harvested, no recorded run)
hlld/cross-iteration-batch_normalization/cifar10.py
code served (permissive licence) · get_code("403a798c6fd52a08")
How it ran

no invocation recorded.

load_pickle Not yet run
unverified (harvested, no recorded run)
hlld/cross-iteration-batch_normalization/cifar10.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from miccaiif/ins, is pointer only (licence: NONE) · get_code("c0233f1d95c36390")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from miccaiif/ins, 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.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 2 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 4 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
load_CIFAR10 1 other paper
1 dated earlier
1702.08835 (2017-02)
Deep Forest
load_pickle 1 other paper
1 later
2307.02249 (2023-07)
Rethinking Multiple Instance Learning for Whole Slide Image Classification: A Good Instance Classifier is All You Need

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 2026-11-21.

Repositories linked to this paper

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

Abstract

A well-known issue of Batch Normalization is its significantly reduced effectiveness in the case of small mini-batch sizes. When a mini-batch contains few examples, the statistics upon which the normalization is defined cannot be reliably estimated from it during a training iteration. To address this problem, we present Cross-Iteration Batch Normalization (CBN), in which examples from multiple recent iterations are jointly utilized to enhance estimation quality. A challenge of computing statistics over multiple iterations is that the network activations from different iterations are not comparable to each other due to changes in network weights. We thus compensate for the network weight changes via a proposed technique based on Taylor polynomials, so that the statistics can be accurately estimated and batch normalization can be effectively applied. On object detection and image classification with small mini-batch sizes, CBN is found to outperform the original batch normalization and a direct calculation of statistics over previous iterations without the proposed compensation technique. Code is available at https://github.com/Howal/Cross-iterationBatchNorm .

For agents

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

get_harvested_code_for_paper("2002.05712")
get_code_for_paper("2002.05712")
have("2002.05712")

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

curl https://syntology.ai/api/ran/2002.05712.json

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

[![Syntology run record](https://syntology.ai/api/ran/2002.05712.svg)](https://syntology.ai/paper/2002.05712)

Connect an agent — have() is free.