SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1609.04836 · 2016

On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima

Jorge Nocedal, Nitish Shirish Keskar, Dheevatsa Mudigere, Mikhail Smelyanskiy, Ping Tak, Peter Tang

arXiv · PDF · Open in the Atlas

Code that ran

15 of 16 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 7 have no contract verdict; 8 failures of our instrument; 1 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/1609.04836.json · badge: /api/ran/1609.04836.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
keskarnitish/large-batch-training canonical (classifier label)
classifier basis: census
3 ran · 0 no recorded run
bedssys/Bedssys community (classifier label)
classifier basis not recorded on the link
12 ran · 1 no recorded run
FunctionStatusWhere it lives
IOU Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bedssys/Bedssys/darknet/gen_anchors.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from caihaocong/Yolov3_AlexeyAB_Version, is pointer only (licence: NOASSERTION) · get_code("ff92ad8f00e93bbc")
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 caihaocong/Yolov3_AlexeyAB_Version, 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.

avg_IOU Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bedssys/Bedssys/darknet/gen_anchors.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from caihaocong/Yolov3_AlexeyAB_Version, is pointer only (licence: NOASSERTION) · get_code("1a3fa3dda9a8b100")
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 caihaocong/Yolov3_AlexeyAB_Version, 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.

c_array Ran
ran; our contract draft was wrong, not the code
bedssys/Bedssys/darknet/darknet.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from caihaocong/Yolov3_AlexeyAB_Version, is pointer only (licence: NOASSERTION) · get_code("78d482d2a7373f3c")
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 caihaocong/Yolov3_AlexeyAB_Version, 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.

convertBack Ran
ran on a synthesized input
bedssys/Bedssys/darknet/darknet_video.py
code served (permissive licence) · get_code("410aa729a6034293")
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 int; 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

torch.manual_seed(0)

ARGS = {
    "x": 100,
    "y": 200,
    "w": 50,
    "h": 75
}
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, ndim 1, dtype family int — 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
cvDrawBoxes Ran
ran on a synthesized input
bedssys/Bedssys/darknet/darknet_video.py
code served (permissive licence) · get_code("0c143d1141be03d4")
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 ndarray, ndim 3, dtype family int, shape (100, 150, 3) 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 cv2
import numpy as np

# Create a small test image (height=100, width=150, channels=3)
img = np.zeros((100, 150, 3), dtype=np.uint8)

# Create detections list with format expected by cvDrawBoxes
# Each detection is a tuple/list: (class_name_bytes, confidence, bbox)
# where bbox = (x_center, y_center, width, height)
detections = [
    (b'person', 0.95, (75, 50, 40, 60)),
    (b'car', 0.87, (100, 30, 50, 35)),
]

ARGS = {
    'detections': detections,
    'img': img,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (100, 150, 3) — shape (100, 150, 3) 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
get_voc_results_file_template Ran
ran on a synthesized input
bedssys/Bedssys/darknet/reval_voc_py3.py
code served (permissive licence) · get_code("79910d76d3273883")
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 os
import torch

ARGS = {
    'image_set': 'val',
    'out_dir': 'results'
}
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 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
make_layers Ran
ran; our contract draft was wrong, not the code
keskarnitish/large-batch-training/PyTorch/vgg.py
code served (permissive licence) · get_code("ac62432dc5134b0d")
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 Sequential. 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)

# The function takes two parameters: cfg and batch_norm
# cfg should be a list/sequence of either integers (channel counts) or 'M' (maxpool marker)
# batch_norm is a boolean flag

# Using one of the predefined configurations from the module-level cfg dict
cfg_list = [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M']

ARGS = {
    'cfg': cfg_list,
    'batch_norm': False
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
parse_rec Ran
ran on a synthesized input
bedssys/Bedssys/darknet/voc_eval_py3.py
code served (permissive licence) · get_code("f58e7859fdb905b4")
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 xml.etree.ElementTree as ET
import tempfile
import os

# Create a temporary PASCAL VOC XML file for testing
xml_content = """<?xml version="1.0" encoding="UTF-8"?>
<annotation>
    <object>
        <name>cat</name>
        <difficult>0</difficult>
        <bndbox>
            <xmin>10</xmin>
            <ymin>20</ymin>
            <xmax>100</xmax>
            <ymax>150</ymax>
        </bndbox>
    </object>
    <object>
        <name>dog</name>
        <difficult>1</difficult>
        <bndbox>
            <xmin>50</xmin>
            <ymin>60</ymin>
            <xmax>200</xmax>
            <ymax>250</ymax>
        </bndbox>
    </object>
</annotation>
"""

# Write XML to temporary file
temp_fd, temp_path = tempfile.mkstemp(suffix='.xml')
try:
    os.write(temp_fd, xml_content.encode('utf-8'))
    os.close(temp_fd)
    
    ARGS = {
        'filename': temp_path
    }
finally:
    pass
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
round_int Ran
ran on a synthesized input
bedssys/Bedssys/image_pose.py
code served (permissive licence) · get_code("feedb4ca621a0244")
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 float, 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

def round_int(val):
    return round(val, 3)

# The function takes a single parameter 'val' and rounds it to 3 decimal places
# val should be a numeric value (int or float)

ARGS = {
    "val": 3.14159
}
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 float, 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
sample Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bedssys/Bedssys/darknet/darknet.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from caihaocong/Yolov3_AlexeyAB_Version, is pointer only (licence: NOASSERTION) · get_code("9bca95aee23d1241")
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 caihaocong/Yolov3_AlexeyAB_Version, 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.

vgg11 Ran
ran; our contract draft was wrong, not the code
keskarnitish/large-batch-training/PyTorch/vgg.py
code served (permissive licence) · get_code("42754302076e329d")
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 VGG. 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
import math

torch.manual_seed(0)

# The function vgg11 takes parameters: pretrained and **kwargs
# Looking at the function body:
# - pretrained: bool, controls whether to load pretrained weights
# - **kwargs: passed to VGG constructor, but VGG.__init__ only takes 'features'
#   so kwargs should be empty or we can pass nothing extra

ARGS = {
    'pretrained': False,
}
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 VGG — a VGG output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VGG 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
vgg13 Ran
ran; our contract draft was wrong, not the code
keskarnitish/large-batch-training/PyTorch/vgg.py
code served (permissive licence) · get_code("2ad86edffbcbccba")
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 VGG. 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
import math

torch.manual_seed(0)

# The function signature is: vgg13(pretrained=False, **kwargs)
# We need to provide values for 'pretrained' and any kwargs

ARGS = {
    'pretrained': False,
}
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 VGG — a VGG output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VGG 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
voc_ap Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
bedssys/Bedssys/darknet/voc_eval_py3.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from SupreethN/Hand-CNN, is pointer only (licence: NOASSERTION) · get_code("95c31c78be42f723")
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 SupreethN/Hand-CNN, 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.

voc_eval Ran
ran on a synthesized input
bedssys/Bedssys/darknet/voc_eval_py3.py
code served (permissive licence) · get_code("5516237bb80c2b46")
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 os
import tempfile
import xml.etree.ElementTree as ET

# Create temporary directory for test files
tmpdir = tempfile.mkdtemp()

# Create a sample annotation XML file
anno_dir = os.path.join(tmpdir, 'annotations')
os.makedirs(anno_dir, exist_ok=True)

# Create sample XML annotation
xml_content = """<?xml version="1.0" encoding="UTF-8"?>
<annotation>
  <object>
    <name>dog</name>
    <difficult>0</difficult>
    <bndbox>
      <xmin>10</xmin>
      <ymin>20</ymin>
      <xmax>100</xmax>
      <ymax>150</ymax>
    </bndbox>
  </object>
  <object>
    <name>cat</name>
    <difficult>0</difficult>
    <bndbox>
      <xmin>50</xmin>
      <ymin>60</ymin>
      <xmax>200</xmax>
      <ymax>250</ymax>
    </bndbox>
  </object>
</annotation>
"""

xml_file1 = os.path.join(anno_dir, 'image001.xml')
with open(xml_file1, 'w') as f:
    f.write(xml_content)

xml_file2 = os.path.join(anno_dir, 'image002.xml')
with open(xml_file2, 'w') as f:
    f.write(xml_content)

# Create imageset file
imageset_file = os.path.join(tmpdir, 'imageset.txt')
with open(imageset_file, 'w') as f:
    f.write('image001\n')
    f.write('image002\n')

# Create detection results file
det_dir = os.path.join(tmpdir, 'detections')
os.makedirs(det_dir, exist_ok=True)

det_file = os.path.join(det_dir, 'comp1_det_dog.txt')
with open(det_file, 'w') as f:
    f.write('image001 0.9 10 20 100 150\n')
    f.write('image002 0.8 50 60 200 250\n')

# Create cache directory
cache_dir = os.path.join(tmpdir, 'cache')
os.makedirs(cache_dir, exist_ok=True)

ARGS = {
    'detpath': os.path.join(det_dir, 'comp1_det_{}.txt'),
    'annopath': os.path.join(anno_dir, '{}.xml'),
    'imagesetfile': imageset_file,
    'classname': 'dog',
    'cachedir': cache_dir,
    'ovthresh': 0.5,
    'use_07_metric': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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
write_coco_json Ran
ran on a synthesized input
bedssys/Bedssys/image_pose.py
code served (permissive licence) · get_code("9535d0a6fc7fd153")
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.

class BodyPart:
    def __init__(self, x, y):
        self.x = x
        self.y = y

class Human:
    def __init__(self):
        self.body_parts = {}

human = Human()
human.body_parts[0] = BodyPart(0.1, 0.2)
human.body_parts[1] = BodyPart(0.3, 0.4)
human.body_parts[2] = BodyPart(0.5, 0.6)
human.body_parts[5] = BodyPart(0.7, 0.8)
human.body_parts[10] = BodyPart(0.9, 0.15)
human.body_parts[15] = BodyPart(0.25, 0.35)

image_w = 640
image_h = 480

ARGS = {
    "human": human,
    "image_w": image_w,
    "image_h": image_h,
}
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
network_width Not yet run
unverified (harvested, no recorded run)
bedssys/Bedssys/darknet/darknet.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from caihaocong/Yolov3_AlexeyAB_Version, is pointer only (licence: NOASSERTION) · get_code("b78a5e60fc79a562")
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 caihaocong/Yolov3_AlexeyAB_Version, 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. No invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 11 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 1 the same month, 11 later) and 5 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
make_layers 59 other papers
3 dated earlier, 56 later
1411.4038 (2014-11)
Fully Convolutional Networks for Semantic Segmentation
1511.00561 (2015-11)
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation
1511.05950 (2015-11)
Staleness-aware Async-SGD for Distributed Deep Learning
and 56 more not named here
voc_ap 27 other papers
2 dated earlier, 1 the same month, 24 later
1506.01497 (2015-06)
Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
1605.06409 (2016-05)
R-FCN: Object Detection via Region-based Fully Convolutional Networks
1609.03605 (2016-09)
Detecting Text in Natural Image with Connectionist Text Proposal Network
and 24 more not named here
c_array 4 other papers
4 later
1612.08242 (2016-12)
YOLO9000: Better, Faster, Stronger
1804.02767 (2018-04)
YOLOv3: An Incremental Improvement
1907.10890 (2019-07)
DeFog: Fog Computing Benchmarks
and 1 more not named here
sample 4 other papers
4 later
1612.08242 (2016-12)
YOLO9000: Better, Faster, Stronger
1804.02767 (2018-04)
YOLOv3: An Incremental Improvement
1907.10890 (2019-07)
DeFog: Fog Computing Benchmarks
and 1 more not named here
IOU 3 other papers
1 dated earlier, 2 later
1602.02282 (2016-02)
Ladder Variational Autoencoders
1612.08242 (2016-12)
YOLO9000: Better, Faster, Stronger
1804.02767 (2018-04)
YOLOv3: An Incremental Improvement
avg_IOU 3 other papers
1 dated earlier, 2 later
1602.02282 (2016-02)
Ladder Variational Autoencoders
1612.08242 (2016-12)
YOLO9000: Better, Faster, Stronger
1804.02767 (2018-04)
YOLOv3: An Incremental Improvement
network_width 3 other papers
3 later
1612.08242 (2016-12)
YOLO9000: Better, Faster, Stronger
1804.02767 (2018-04)
YOLOv3: An Incremental Improvement
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
convertBack 1 other paper
1 later
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
cvDrawBoxes 1 other paper
1 later
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
vgg11 1 other paper
1 later
1805.11614 (2018-05)
Deep Learning under Privileged Information Using Heteroscedastic Dropout
vgg13 1 other paper
1 later
1805.11614 (2018-05)
Deep Learning under Privileged Information Using Heteroscedastic Dropout

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

The stochastic gradient descent (SGD) method and its variants are algorithms of choice for many Deep Learning tasks. These methods operate in a small-batch regime wherein a fraction of the training data, say 32-512 data points, is sampled to compute an approximation to the gradient. It has been observed in practice that when using a larger batch there is a degradation in the quality of the model, as measured by its ability to generalize. We investigate the cause for this generalization drop in the large-batch regime and present numerical evidence that supports the view that large-batch methods tend to converge to sharp minimizers of the training and testing functions-and as is well known, sharp minima lead to poorer generalization. In contrast, small-batch methods consistently converge to flat minimizers, and our experiments support a commonly held view that this is due to the inherent noise in the gradient estimation. We discuss several strategies to attempt to help large-batch methods eliminate this generalization gap.

For agents

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

get_harvested_code_for_paper("1609.04836")
get_code_for_paper("1609.04836")
have("1609.04836")

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

curl https://syntology.ai/api/ran/1609.04836.json

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

[![Syntology run record](https://syntology.ai/api/ran/1609.04836.svg)](https://syntology.ai/paper/1609.04836)

Connect an agent — have() is free.