SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.11930 · ICML · 2021

On the importance of cross-task features for class-incremental learning

Marc Masana, Bartłomiej Twardowski, Joost Van De Weijer, Albin Soutif-Cormerais

arXiv · PDF · Open in the Atlas

Code that ran

3 of 5 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 3 have no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-22; 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/2106.11930.json · badge: /api/ran/2106.11930.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
AlbinSou/cross_task_cil canonical (classifier label)
classifier basis: census
3 ran · 2 no recorded run
FunctionStatusWhere it lives
extract_forgetting Ran
ran on a synthesized input
AlbinSou/cross_task_cil/src/cumulative_metrics.py
code served (permissive licence) · get_code("dc430dd8ca7f26e9")
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 2, dtype family float, shape (4, 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 numpy as np

# The function extract_forgetting takes accs as input
# accs should be a 2D numpy array where:
# - First dimension represents tasks being trained on
# - Second dimension represents tasks being evaluated on
# The function accesses accs with [task_train, task_eval] and [:task_train + 1, task_eval]
# So we need a 2D array with accuracy values

# Create a small accuracy matrix: 4 tasks, 3 evaluation metrics
# Each entry represents accuracy of task_train on task_eval
accs = np.array([
    [0.95, 0.80, 0.70],
    [0.92, 0.85, 0.72],
    [0.90, 0.83, 0.75],
    [0.88, 0.81, 0.78]
], dtype=np.float32)

ARGS = {
    "accs": accs
}
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 ndarray, ndim 2, dtype family float, shape (4, 3) — shape (4, 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_data Ran
ran on a synthesized input
AlbinSou/cross_task_cil/src/datasets/base_dataset.py
code served (permissive licence) · get_code("544cf571f3a07cc3")
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 numpy as np

# Create temporary directory with required data files
tmpdir = tempfile.mkdtemp()

# Create train.txt file with image paths and labels
train_data = [
    "img_0.jpg 0",
    "img_1.jpg 1",
    "img_2.jpg 2",
    "img_3.jpg 3",
    "img_4.jpg 0",
    "img_5.jpg 1",
    "img_6.jpg 2",
    "img_7.jpg 3",
]
train_file = os.path.join(tmpdir, 'train.txt')
with open(train_file, 'w') as f:
    f.write('\n'.join(train_data))

# Create test.txt file with image paths and labels
test_data = [
    "img_8.jpg 0",
    "img_9.jpg 1",
    "img_10.jpg 2",
    "img_11.jpg 3",
]
test_file = os.path.join(tmpdir, 'test.txt')
with open(test_file, 'w') as f:
    f.write('\n'.join(test_data))

# Create dummy image files so os.path.isabs check works
for i in range(12):
    img_path = os.path.join(tmpdir, f'img_{i}.jpg')
    with open(img_path, 'w') as f:
        f.write('dummy')

ARGS = {
    'path': tmpdir,
    'num_tasks': 2,
    'nc_first_task': 2,
    'validation': 0.25,
    'shuffle_classes': False,
    'class_order': None,
}
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
get_transforms Ran
ran on a synthesized input
AlbinSou/cross_task_cil/src/datasets/data_loader.py
code served (permissive licence) · get_code("7210d296e5161f2a")
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 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 torchvision.transforms as transforms

torch.manual_seed(0)

ARGS = {
    "resize": (224, 224),
    "pad": 10,
    "crop": 200,
    "flip": True,
    "normalize": ([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
    "extend_channel": 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — 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 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
get_datasets Not yet run
unverified (harvested, no recorded run)
AlbinSou/cross_task_cil/src/datasets/data_loader.py
code served (permissive licence) · get_code("2125ab30557c552f")
How it ran

no invocation recorded.

get_loaders Not yet run
unverified (harvested, no recorded run)
AlbinSou/cross_task_cil/src/datasets/data_loader.py
code served (permissive licence) · get_code("00e4bb5fb503e035")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 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.

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

In class-incremental learning, an agent with limited resources needs to learn a sequence of classification tasks, forming an ever growing classification problem, with the constraint of not being able to access data from previous tasks. The main difference with task-incremental learning, where a task-ID is available at inference time, is that the learner also needs to perform cross-task discrimination, i.e. distinguish between classes that have not been seen together. Approaches to tackle this problem are numerous and mostly make use of an external memory (buffer) of nonnegligible size. In this paper, we ablate the learning of cross-task features and study its influence on the performance of basic replay strategies used for class-IL. We also define a new forgetting measure for class-incremental learning, and see that forgetting is not the principal cause of low performance. Our experimental results show that future algorithms for class-incremental learning should not only prevent forgetting, but also aim to improve the quality of the cross-task features, and the knowledge transfer between tasks. This is especially important when tasks contain limited amount of data.

For agents

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

get_harvested_code_for_paper("2106.11930")
get_code_for_paper("2106.11930")
have("2106.11930")

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

curl https://syntology.ai/api/ran/2106.11930.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.11930.svg)](https://syntology.ai/paper/2106.11930)

Connect an agent — have() is free.