SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2009.08107 · NeurIPS · 2020

Few-Shot Unsupervised Continual Learning through Meta-Examples

Simone Calderara, Andrea Passerini, Alessia Bertugli, Stefano Vincenzi

arXiv · PDF · Open in the Atlas

Code that ran

6 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 1 failure of our instrument; 2 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/2009.08107.json · badge: /api/ran/2009.08107.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
alessiabertugli/FUSION-ME canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
alessiabertugli/FUSION marked official in Papers with Code
Papers with Code archive
2 ran · 2 no recorded run
FunctionStatusWhere it lives
ReservoirSampler Ran
ran on a synthesized input · constructed an object; no result was computed
alessiabertugli/FUSION/model/meta_learner.py
code served (permissive licence) · get_code("ba43569331dc53d8")
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 an instance of its own class, ReservoirSampler: the run constructed the object and did not call it, so no result was computed. 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 random
import torch

torch.manual_seed(0)

# ReservoirSampler expects:
# - windows: an integer representing the window size for sampling
# - buffer_size: an integer for the maximum buffer size (default 5000)

# Looking at the usage in the class:
# - windows is used in: random.randint(0, min(self.total_additions, self.window))
# - buffer_size is used to limit buffer length and modulo operation
# - add() takes data which is stored in a list
# - sample() calls random.sample on the buffer
# - sample_trajectory() slices the buffer

# We'll create small sample data items (tensors work fine as list elements)
sample_data_1 = torch.randn(2, 4)
sample_data_2 = torch.randn(2, 4)
sample_data_3 = torch.randn(2, 4)

ARGS = {
    "windows": 100,
    "buffer_size": 50
}
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 ReservoirSampler — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ReservoirSampler 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
batch_norm Ran
ran on a synthesized input
alessiabertugli/FUSION-ME/model/layers.py
code served (permissive licence) · get_code("9efa468ab93d1d7a")
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 torch
from torch import nn

torch.manual_seed(0)

ARGS = {
    'param': {
        'in-channels': 4
    }
}
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
conv2d Ran
ran on a synthesized input
alessiabertugli/FUSION-ME/model/layers.py
code served (permissive licence) · get_code("ad6ffadd1eb45101")
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 torch
from torch import nn

torch.manual_seed(0)

# Build param dict with required keys
param = {
    'out-channels': 16,
    'in-channels': 3,
    'kernal': 5
}

# Build adaptation - used as assignment target, just needs to be a value
adaptation = "adaptation_value"

# Build meta - used as assignment target, just needs to be a value
meta = "meta_value"

ARGS = {
    'param': param,
    'adaptation': adaptation,
    'meta': meta
}
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
find_index Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
alessiabertugli/FUSION/model/meta_learner.py
code served (permissive licence) · get_code("9573e1d8cdd1e4b8")
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 int, ndim 0, dtype family int, 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 numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - find_index takes y_traj and y_rand
# - It accesses y_traj[0] and y_rand[0]
# - It enumerates over y_rand[0], getting idx and label
# - It compares label == y_traj[0]
# - It returns idx when a match is found

# y_rand appears to be a list/tensor where y_rand[0] is iterable
# y_traj appears to be a list/tensor where y_traj[0] is a single value to compare

# Create small tensors to match the usage pattern
y_traj = torch.tensor([2, 5, 7, 1])  # y_traj[0] = 2
y_rand = torch.tensor([[2, 3, 4, 5], [6, 7, 8, 9]])  # y_rand[0] = [2, 3, 4, 5]

# Verify the logic: y_rand[0][0] should equal y_traj[0] to return 0
assert y_traj[0].item() == y_rand[0][0].item(), "Setup verification"

ARGS = {
    "y_traj": y_traj,
    "y_rand": y_rand,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, 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
get_model Ran
ran on a synthesized input
alessiabertugli/FUSION-ME/model/modelfactory.py
code served (permissive licence) · get_code("433119ef21014064")
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 torch
import torch.nn as nn

class SimpleArgs:
    def __init__(self, dataset, attention=False, mean=False):
        self.dataset = dataset
        self.attention = attention
        self.mean = mean

ARGS = {
    'args': SimpleArgs(dataset='cifar', attention=False, mean=False),
    'classes': 10
}
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
linear Ran
ran on a synthesized input
alessiabertugli/FUSION-ME/model/layers.py
code served (permissive licence) · get_code("b0e5044d51d9190f")
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 torch
from torch import nn

torch.manual_seed(0)

# Analyze the function:
# - out_dim: int (used as first dimension of weight tensor)
# - in_dim: int (used as second dimension of weight tensor)
# - adaptation: object (assigned to w.adaptation and b.adaptation attributes)
# - meta: object (assigned to w.meta and b.meta attributes)

# Create simple stand-in objects for adaptation and meta
class AdaptationConfig:
    def __init__(self):
        self.enabled = True

class MetaConfig:
    def __init__(self):
        self.version = 1

ARGS = {
    "out_dim": 4,
    "in_dim": 3,
    "adaptation": AdaptationConfig(),
    "meta": MetaConfig(),
}
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
Learner Not yet run
unverified (harvested, no recorded run)
alessiabertugli/FUSION/model/meta_learner.py
code served (permissive licence) · get_code("93eab00ecefb419a")
How it ran

no invocation recorded.

MetaLearnerClassification Not yet run
unverified (harvested, no recorded run)
alessiabertugli/FUSION/model/meta_learner.py
code served (permissive licence) · get_code("dcdffd1e41cc9162")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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 8 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 real-world applications, data do not reflect the ones commonly used for neural networks training, since they are usually few, unlabeled and can be available as a stream. Hence many existing deep learning solutions suffer from a limited range of applications, in particular in the case of online streaming data that evolve over time. To narrow this gap, in this work we introduce a novel and complex setting involving unsupervised meta-continual learning with unbalanced tasks. These tasks are built through a clustering procedure applied to a fitted embedding space. We exploit a meta-learning scheme that simultaneously alleviates catastrophic forgetting and favors the generalization to new tasks. Moreover, to encourage feature reuse during the meta-optimization, we exploit a single inner loop taking advantage of an aggregated representation achieved through the use of a selfattention mechanism. Experimental results on few-shot learning benchmarks show competitive performance even compared to the supervised case. Additionally, we empirically observe that in an unsupervised scenario, the small tasks and the variability in the clusters pooling play a crucial role in the generalization capability of the network. Further, on complex datasets, the exploitation of more clusters than the true number of classes leads to higher results, even compared to the ones obtained with full supervision, suggesting that a predefined partitioning into classes can miss relevant structural information. The code is available at https://github.com/alessiabertugli/FUSION-ME

For agents

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

get_harvested_code_for_paper("2009.08107")
get_code_for_paper("2009.08107")
have("2009.08107")

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

curl https://syntology.ai/api/ran/2009.08107.json

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

[![Syntology run record](https://syntology.ai/api/ran/2009.08107.svg)](https://syntology.ai/paper/2009.08107)

Connect an agent — have() is free.