SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2012.03107 · ICLR · 2021

When Do Curricula Work?

Behnam Neyshabur, Xiaoxia Wu, Ethan Dyer

arXiv · PDF · Open in the Atlas

Code that ran

2 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 2 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/2012.03107.json · badge: /api/ran/2012.03107.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
google-research/understanding-curricula canonical (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
0 ran · 1 no recorded run
FunctionStatusWhere it lives
cuda_transfer Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
google-research/understanding-curricula/loss-based-cscore.py
code served (permissive licence) · get_code("b169c4c59c6e6c2b")
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. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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)

# Create small CPU tensors
images = torch.randn(2, 3, 4, 4)  # batch=2, channels=3, height=4, width=4
target = torch.randint(0, 10, (2,))  # batch=2, class labels

ARGS = {
    "images": images,
    "target": target,
}
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
validate Ran
ran; our contract draft was wrong, not the code
google-research/understanding-curricula/loss-based-cscore.py
code served (permissive licence) · get_code("4dceb83470c09763")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 torch.utils.data

torch.manual_seed(0)

# Small model that takes images and outputs logits for classification
class SimpleModel(nn.Module):
    def __init__(self, input_dim=4, num_classes=3):
        super().__init__()
        self.fc = nn.Linear(input_dim, num_classes)
    
    def forward(self, x):
        return self.fc(x)

# Create a small dataset with 4 samples
class SimpleDataset:
    def __init__(self):
        self.data = [
            (torch.randn(1, 4), torch.tensor(0)),
            (torch.randn(1, 4), torch.tensor(1)),
            (torch.randn(1, 4), torch.tensor(2)),
            (torch.randn(1, 4), torch.tensor(1)),
        ]
    
    def __iter__(self):
        return iter(self.data)
    
    def __len__(self):
        return len(self.data)

# Build val_loader as a list of (images, target) tuples
# Each batch has 2 samples, each image is shape (1, 4), targets are scalars
val_loader = [
    (torch.randn(2, 4), torch.tensor([0, 1])),
    (torch.randn(2, 4), torch.tensor([2, 1])),
]

# Build model
model = SimpleModel(input_dim=4, num_classes=3)

# Build criterion (CrossEntropyLoss returns per-sample losses when reduction='none')
criterion = nn.CrossEntropyLoss(reduction='none')

# Build val_order - list of indices mapping samples to positions
val_order = [0, 1, 2, 3]

# Build ind_loss - dict mapping sample indices to lists of losses
ind_loss = {0: [], 1: [], 2: [], 3: []}

ARGS = {
    'val_loader': val_loader,
    'model': model,
    'criterion': criterion,
    'val_order': val_order,
    'ind_loss': ind_loss,
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cuda_transfer Not yet run
unverified (harvested, no recorded run)
this paper's copy was not recorded; identical code first harvested from pengbohua/angulargap
pointer only; the stored copy get_code serves, from pengbohua/angulargap, is pointer only (licence: NONE) · get_code("9b4b3a342500d694")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 2 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 2 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

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

FunctionAlso present inEarliest-datedOthers named
cuda_transfer 1 other paper
1 later
2401.02718 (2024-01)
Calibration Attacks: A Comprehensive Study of Adversarial Attacks on Model Confidence
cuda_transfer 1 other paper
1 later
2207.08525 (2022-07)
Angular Gap: Reducing the Uncertainty of Image Difficulty through Model Calibration

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

Inspired by human learning, researchers have proposed ordering examples during training based on their difficulty. Both curriculum learning, exposing a network to easier examples early in training, and anti-curriculum learning, showing the most difficult examples first, have been suggested as improvements to the standard i.i.d. training. In this work, we set out to investigate the relative benefits of ordered learning. We first investigate the implicit curricula resulting from architectural and optimization bias and find that samples are learned in a highly consistent order. Next, to quantify the benefit of explicit curricula, we conduct extensive experiments over thousands of orderings spanning three kinds of learning: curriculum, anti-curriculum, and random-curriculum -in which the size of the training dataset is dynamically increased over time, but the examples are randomly ordered. We find that for standard benchmark datasets, curricula have only marginal benefits, and that randomly ordered samples perform as well or better than curricula and anti-curricula, suggesting that any benefit is entirely due to the dynamic training set size. Inspired by common use cases of curriculum learning in practice, we investigate the role of limited training time budget and noisy data in the success of curriculum learning. Our experiments demonstrate that curriculum, but not anti-curriculum can indeed improve the performance either with limited training time budget or in existence of noisy data. * Work performed in part while Xiaoxia Wu was interning at Google. 1 Both GPT-3 and T5 have non-uniform mixing strategies. See (Brown et al., 2020, Section 2.2, Table 2.2) and (Raffel et al., 2019, Section 3.5) for more details.

For agents

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

get_harvested_code_for_paper("2012.03107")
get_code_for_paper("2012.03107")
have("2012.03107")

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

curl https://syntology.ai/api/ran/2012.03107.json

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

[![Syntology run record](https://syntology.ai/api/ran/2012.03107.svg)](https://syntology.ai/paper/2012.03107)

Connect an agent — have() is free.