SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2003.13042 · 2020

Omni-sourced Webly-supervised Learning for Video Recognition

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 9 have no contract verdict; 1 failure of our instrument; 3 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/2003.13042.json · badge: /api/ran/2003.13042.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
open-mmlab/mmaction canonical (classifier label)
classifier basis: census
10 ran · 3 no recorded run
FunctionStatusWhere it lives
SSIM_loss Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/losses/flow_losses.py
code served (permissive licence) · get_code("4bed2534f3a2204f")
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 Tensor, 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
import math

torch.manual_seed(0)

# Create small image tensors: batch_size=2, channels=3, height=16, width=16
img1 = torch.randn(2, 3, 16, 16)
img2 = torch.randn(2, 3, 16, 16)

# Define all parameters
ARGS = {
    'img1': img1,
    'img2': img2,
    'kernel_size': 8,
    'stride': 8,
    'c1': 1e-05,
    'c2': 1e-05
}
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 Tensor, 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
bbox_overlaps Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
open-mmlab/mmaction/mmaction/core/bbox2d/geometry.py
code served (permissive licence) · get_code("d65ea05c493bc2d8")
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 Tensor, ndim 2, dtype family float, shape (2, 2) 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

torch.manual_seed(0)

# Create two sets of bounding boxes
# Format: (x1, y1, x2, y2) where x1, y1 is top-left and x2, y2 is bottom-right
bboxes1 = torch.tensor([
    [10.0, 10.0, 50.0, 50.0],
    [20.0, 20.0, 60.0, 60.0]
], dtype=torch.float32)

bboxes2 = torch.tensor([
    [15.0, 15.0, 55.0, 55.0],
    [30.0, 30.0, 70.0, 70.0]
], dtype=torch.float32)

ARGS = {
    'bboxes1': bboxes1,
    'bboxes2': bboxes2,
    'mode': 'iou',
    'is_aligned': 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 Tensor, ndim 2, dtype family float, shape (2, 2) — shape (2, 2) 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
charbonnier_loss Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/losses/flow_losses.py
code served (permissive licence) · get_code("d1055ad5929a8347")
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 Tensor, 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

torch.manual_seed(0)

# Create a small difference tensor (batch=2, channels=4, height=3, width=5)
difference = torch.randn(2, 4, 3, 5)

# Create a mask tensor with matching spatial dimensions
# mask should have shape (batch=2, 1 or channels, height=3, width=5)
mask = torch.ones(2, 1, 3, 5)

# Set alpha and beta parameters
alpha = 1
beta = 1.0
epsilon = 0.001

ARGS = {
    "difference": difference,
    "mask": mask,
    "alpha": alpha,
    "beta": beta,
    "epsilon": epsilon,
}
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 Tensor, 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
classwise_regression_loss Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/losses/ssn_losses.py
code served (permissive licence) · get_code("d05f21f2da402e1c")
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 Tensor, 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.functional as F

torch.manual_seed(0)

# Analyze the function:
# - pred: indexed as pred[:, indexer, :] where indexer comes from labels.data - 1
#   So pred needs shape (batch, num_classes, 2) based on the diag operations
# - labels: used as labels.data - 1, so needs to be a tensor with integer values
# - targets: used in smooth_l1_loss, needs to match the shape of class_pred which is (batch, 2)

batch_size = 2
num_classes = 4
feature_dim = 2

# pred: (batch_size, num_classes, feature_dim)
# The function does prep = pred[:, indexer, :] where indexer is from labels
# Then extracts diagonals from prep[:, :, 0] and prep[:, :, 1]
# This means pred should have shape (batch, num_classes, 2)
pred = torch.randn(batch_size, num_classes, feature_dim)

# labels: used as labels.data - 1 to create indexer
# indexer should select from num_classes dimension, so values should be in range [1, num_classes]
labels = torch.tensor([1, 2], dtype=torch.long)

# targets: compared with class_pred which has shape (batch, 2)
# smooth_l1_loss will flatten both, so targets should have compatible shape
targets = torch.randn(batch_size, 2)

ARGS = {
    "pred": pred,
    "labels": labels,
    "targets": targets,
}
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 Tensor, 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
expand_target Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/core/bbox2d/bbox_target.py
code served (permissive licence) · get_code("a57ba57f6dde621b")
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 3, 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

torch.manual_seed(0)

# Create small tensors with batch_size=2, bbox_dim=4
batch_size = 2
num_classes = 3

# bbox_targets: (batch_size, 4) - bounding box regression targets
bbox_targets = torch.randn(batch_size, 4)

# bbox_weights: (batch_size, 4) - weights for each bbox coordinate
bbox_weights = torch.ones(batch_size, 4)

# labels: (batch_size,) - class labels, values should be > 0 for some samples
# Using values in range [0, num_classes) where > 0 means object is present
labels = torch.tensor([0, 1], dtype=torch.long)

ARGS = {
    "bbox_targets": bbox_targets,
    "bbox_weights": bbox_weights,
    "labels": labels,
    "num_classes": num_classes,
}
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 3, 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
images_to_levels Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/core/anchor2d/anchor_target.py
code served (permissive licence) · get_code("8fa249eba3a0ce57")
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. 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

torch.manual_seed(0)

# The function expects:
# - target: a list of tensors (one per image)
# - num_level_anchors: a list of integers representing anchor counts per level

# Create a list of target tensors (one per image)
# Each tensor has shape (num_anchors,) where num_anchors is the total across all levels
target = [
    torch.randn(7),  # Image 0: 7 total anchors
    torch.randn(7),  # Image 1: 7 total anchors
]

# Define how many anchors per level
# Total anchors = 3 + 4 = 7
num_level_anchors = [3, 4]

ARGS = {
    "target": target,
    "num_level_anchors": num_level_anchors,
}
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 — a list 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
temporal_iou Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/core/bbox1d/geometry.py
code served (permissive licence) · get_code("6fb204c4574e77ac")
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

torch.manual_seed(0)

# temporal_iou takes two temporal spans (start, end tuples)
# Let's create two example spans

span_A = (1.0, 5.0)  # start=1.0, end=5.0
span_B = (3.0, 7.0)  # start=3.0, end=7.0

ARGS = {
    "span_A": span_A,
    "span_B": span_B,
}
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
weighted_binary_cross_entropy Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/losses/losses.py
code served (permissive licence) · get_code("380e288d5962b717")
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 Tensor, ndim 1, dtype family float, shape (1,) 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.functional as F

torch.manual_seed(0)

# Create small tensors for the function parameters
batch_size = 2
num_classes = 4

# pred: predictions with logits, shape (batch_size, num_classes)
pred = torch.randn(batch_size, num_classes)

# label: class indices or binary labels, shape (batch_size,)
# Using values that will trigger the _expand_binary_labels function
label = torch.tensor([1, 3], dtype=torch.long)

# weight: sample weights, shape (batch_size,)
weight = torch.tensor([1.0, 1.0], dtype=torch.float32)

# avg_factor: optional averaging factor
avg_factor = None

ARGS = {
    "pred": pred,
    "label": label,
    "weight": weight,
    "avg_factor": avg_factor
}
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 Tensor, ndim 1, dtype family float, shape (1,) — shape (1,) 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
weighted_cross_entropy Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/losses/losses.py
code served (permissive licence) · get_code("8a052a08051eb8bb")
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 Tensor, ndim 1, dtype family float, shape (1,) 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.functional as F

torch.manual_seed(0)

# Create small tensors for testing
batch_size = 2
num_classes = 4

# pred: logits tensor of shape (batch_size, num_classes)
pred = torch.randn(batch_size, num_classes)

# label: target class indices of shape (batch_size,)
label = torch.tensor([0, 2])

# weight: per-sample weights of shape (batch_size,)
weight = torch.tensor([1.0, 0.5])

# avg_factor: optional average factor (None means it will be computed)
avg_factor = None

# reduce: whether to reduce the loss
reduce = True

ARGS = {
    "pred": pred,
    "label": label,
    "weight": weight,
    "avg_factor": avg_factor,
    "reduce": reduce,
}
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 Tensor, ndim 1, dtype family float, shape (1,) — shape (1,) 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
weighted_nll_loss Ran
ran on a synthesized input
open-mmlab/mmaction/mmaction/losses/losses.py
code served (permissive licence) · get_code("48433fd136b26537")
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 Tensor, ndim 1, dtype family float, shape (1,) 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.functional as F

torch.manual_seed(0)

# pred: input to F.nll_loss, should be log probabilities with shape (batch_size, num_classes)
pred = torch.randn(2, 4).log_softmax(dim=1)

# label: target class indices for F.nll_loss, shape (batch_size,)
label = torch.tensor([0, 2])

# weight: per-sample weights, shape (batch_size,)
weight = torch.tensor([1.0, 0.5])

# avg_factor: optional normalization factor
avg_factor = None

ARGS = {
    'pred': pred,
    'label': label,
    'weight': weight,
    'avg_factor': avg_factor,
}
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 Tensor, ndim 1, dtype family float, shape (1,) — shape (1,) 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
anchor_target_single Not yet run
unverified (harvested, no recorded run)
open-mmlab/mmaction/mmaction/core/anchor2d/anchor_target.py
code served (permissive licence) · get_code("8499456603e86421")
How it ran

no invocation recorded.

bbox_target_single Not yet run
unverified (harvested, no recorded run)
open-mmlab/mmaction/mmaction/core/bbox2d/bbox_target.py
code served (permissive licence) · get_code("bea6a74545a6b1b9")
How it ran

no invocation recorded.

completeness_loss Not yet run
unverified (harvested, no recorded run)
open-mmlab/mmaction/mmaction/losses/ssn_losses.py
code served (permissive licence) · get_code("66a6fb33d71062f2")
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, 7 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 2 the same month, 5 later) and 6 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
images_to_levels 6 other papers
3 dated earlier, 1 the same month, 2 later
1903.00621 (2019-03)
Feature Selective Anchor-Free Module for Single-Shot Object Detection
1906.07155 (2019-06)
MMDetection: Open MMLab Detection Toolbox and Benchmark
1907.03958 (2019-07)
Attentive CT Lesion Detection Using Deep Pyramid Inference with Multi-Scale Booster
and 3 more not named here
anchor_target_single 2 other papers
1 dated earlier, 1 the same month
1903.00621 (2019-03)
Feature Selective Anchor-Free Module for Single-Shot Object Detection
2003.10152 (2020-03)
SOLOv2: Dynamic and Fast Instance Segmentation
bbox_overlaps 2 other papers
2 dated earlier
1906.07155 (2019-06)
MMDetection: Open MMLab Detection Toolbox and Benchmark
1908.09492 (2019-08)
Class-balanced Grouping and Sampling for Point Cloud 3D Object Detection
expand_target 2 other papers
1 dated earlier, 1 later
1906.07155 (2019-06)
MMDetection: Open MMLab Detection Toolbox and Benchmark
2004.06002 (2020-04)
Dynamic R-CNN: Towards High Quality Object Detection via Dynamic Training
weighted_binary_cross_entropy 1 other paper
1 later
2004.03548 (2020-04)
Temporal Pyramid Network for Action Recognition
weighted_cross_entropy 1 other paper
1 later
2004.03548 (2020-04)
Temporal Pyramid Network for Action Recognition
weighted_nll_loss 1 other paper
1 later
2004.03548 (2020-04)
Temporal Pyramid Network for Action Recognition

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

We introduce OmniSource, a novel framework for leveraging web data to train video recognition models. OmniSource overcomes the barriers between data formats, such as images, short videos, and long untrimmed videos for webly-supervised learning. First, data samples with multiple formats, curated by task-specific data collection and automatically filtered by a teacher model, are transformed into a unified form. Then a joint-training strategy is proposed to deal with the domain gaps between multiple data sources and formats in webly-supervised learning. Several good practices, including data balancing, resampling, and cross-dataset mixup are adopted in joint training. Experiments show that by utilizing data from multiple sources and formats, OmniSource is more data-efficient in training. With only 3.5M images and 800K minutes videos crawled from the internet without human labeling (less than 2% of prior works), our models learned with OmniSource improve Top-1 accuracy of 2D- and 3D-ConvNet baseline models by 3.0% and 3.9%, respectively, on the Kinetics-400 benchmark. With OmniSource, we establish new records with different pretraining strategies for video recognition. Our best models achieve 80.4%, 80.5%, and 83.6 Top-1 accuracies on the Kinetics-400 benchmark respectively for training-from-scratch, ImageNet pre-training and IG-65M pre-training.

For agents

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

get_harvested_code_for_paper("2003.13042")
get_code_for_paper("2003.13042")
have("2003.13042")

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

curl https://syntology.ai/api/ran/2003.13042.json

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

[![Syntology run record](https://syntology.ai/api/ran/2003.13042.svg)](https://syntology.ai/paper/2003.13042)

Connect an agent — have() is free.