SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1906.02717 · 2019

Adaptive Gradient-Based Meta-Learning Methods

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 (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 3 have no contract verdict; 2 failures 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/1906.02717.json · badge: /api/ran/1906.02717.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
mkhodak/ARUBA canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
average_vars Ran
ran, violated the contract we drafted
mkhodak/ARUBA/reptile/supervised_reptile/variables.py
code served (permissive licence) · get_code("59c0d53cfcbb2398")
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 2, 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 numpy as np

# Create sample variable sequences
# The function expects var_seqs to be an iterable of sequences
# Each sequence will be zipped together, so we need multiple sequences
# with the same number of elements

# Create 3 sequences, each containing 4 variable arrays
seq1 = [np.array([1.0, 2.0, 3.0]), np.array([4.0, 5.0, 6.0]), 
        np.array([7.0, 8.0, 9.0]), np.array([10.0, 11.0, 12.0])]

seq2 = [np.array([2.0, 3.0, 4.0]), np.array([5.0, 6.0, 7.0]), 
        np.array([8.0, 9.0, 10.0]), np.array([11.0, 12.0, 13.0])]

seq3 = [np.array([3.0, 4.0, 5.0]), np.array([6.0, 7.0, 8.0]), 
        np.array([9.0, 10.0, 11.0]), np.array([12.0, 13.0, 14.0])]

var_seqs = [seq1, seq2, seq3]

ARGS = {
    "var_seqs": var_seqs
}
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 list, ndim 2, 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
interpolate_vars Ran
ran; our contract draft was wrong, not the code
mkhodak/ARUBA/reptile/supervised_reptile/variables.py
code served (permissive licence) · get_code("382960e74b17b40a")
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
import numpy as np

torch.manual_seed(0)

# Create variable sequences as lists of tensors
# These will be used in the interpolate_vars function which performs:
# add_vars(old_vars, scale_vars(subtract_vars(new_vars, old_vars), epsilon))

old_vars = [
    torch.tensor([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]]),
    torch.tensor([[0.5, 1.5, 2.5, 3.5]])
]

new_vars = [
    torch.tensor([[2.0, 3.0, 4.0, 5.0], [6.0, 7.0, 8.0, 9.0]]),
    torch.tensor([[1.5, 2.5, 3.5, 4.5]])
]

epsilon = 0.5

ARGS = {
    "old_vars": old_vars,
    "new_vars": new_vars,
    "epsilon": epsilon
}
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 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
model_kwargs Ran
ran on a synthesized input
mkhodak/ARUBA/reptile/supervised_reptile/args.py
code served (permissive licence) · get_code("856c596403b19ef3")
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. 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

class ParsedArgs:
    def __init__(self):
        self.learning_rate = 0.01
        self.sgd = False
        self.adagrad = False
        self.aruba = True
        self.inner_iters = 5
        self.zeta = 0.1
        self.p = 2
        self.decay = 0.9
        self.adaptive = 'momentum'
        self.train_with_sgd = True

ARGS = {
    'parsed_args': ParsedArgs()
}
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
read_dataset Ran
ran on a synthesized input
mkhodak/ARUBA/reptile/supervised_reptile/miniimagenet.py
code served (permissive licence) · get_code("a58a72b75ca6b4e1")
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 2, 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 os
import random
import tempfile
import shutil
from PIL import Image
import numpy as np

# Create a temporary directory structure that mimics Mini-ImageNet
temp_base = tempfile.mkdtemp()

# Create train, val, test subdirectories
for split in ['train', 'val', 'test']:
    split_dir = os.path.join(temp_base, split)
    os.makedirs(split_dir)
    
    # Create a few WNID class directories (starting with 'n')
    for class_idx in range(2):
        class_dir = os.path.join(split_dir, f'n{class_idx:08d}')
        os.makedirs(class_dir)
        
        # Create a few JPEG images in each class directory
        for img_idx in range(3):
            img_path = os.path.join(class_dir, f'image_{img_idx}.JPEG')
            # Create a small 100x100 RGB image
            img = Image.new('RGB', (100, 100), color=(73, 109, 137))
            img.save(img_path)

ARGS = {
    'data_dir': temp_base
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, 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
split_dataset Ran
ran on a synthesized input
mkhodak/ARUBA/reptile/supervised_reptile/omniglot.py
code served (permissive licence) · get_code("60258f666e8cd77e")
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 random

# Define a minimal Character class to represent dataset elements
class Character:
    def __init__(self, name, value):
        self.name = name
        self.value = value
    
    def __repr__(self):
        return f"Character({self.name}, {self.value})"

# Create a small dataset of Character objects
dataset = [
    Character("a", 1),
    Character("b", 2),
    Character("c", 3),
    Character("d", 4),
    Character("e", 5),
    Character("f", 6),
    Character("g", 7),
    Character("h", 8),
    Character("i", 9),
    Character("j", 10),
    Character("k", 11),
    Character("l", 12),
    Character("m", 13),
    Character("n", 14),
    Character("o", 15),
    Character("p", 16),
    Character("q", 17),
    Character("r", 18),
    Character("s", 19),
    Character("t", 20),
]

ARGS = {
    "dataset": dataset,
    "num_train": 1200
}
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 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
subtract_vars Ran
ran; our contract draft was wrong, not the code
mkhodak/ARUBA/reptile/supervised_reptile/variables.py
code served (permissive licence) · get_code("3880e32be4fd7c7f")
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
import torch.nn as nn

torch.manual_seed(0)

# The function subtract_vars takes two sequences of variables and subtracts
# corresponding elements. Looking at the implementation, it uses:
# - zip() to pair elements
# - subtraction operator (v1 - v2) on each pair
# This works with any objects that support subtraction, but typically tensors.

# Create two sequences of tensors with distinct dimensions for probing
var_seq_1 = [
    torch.randn(2, 4),  # batch 2, dim 4
    torch.randn(2, 3),  # batch 2, dim 3
    torch.randn(2, 5),  # batch 2, dim 5
]

var_seq_2 = [
    torch.randn(2, 4),  # batch 2, dim 4
    torch.randn(2, 3),  # batch 2, dim 3
    torch.randn(2, 5),  # batch 2, dim 5
]

ARGS = {
    "var_seq_1": var_seq_1,
    "var_seq_2": var_seq_2,
}
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
evaluate_kwargs Not yet run
unverified (harvested, no recorded run)
mkhodak/ARUBA/reptile/supervised_reptile/args.py
code served (permissive licence) · get_code("c3a24a121b7eee20")
How it ran

no invocation recorded.

train_kwargs Not yet run
unverified (harvested, no recorded run)
mkhodak/ARUBA/reptile/supervised_reptile/args.py
code served (permissive licence) · get_code("64a961b91d33ed69")
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, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 3 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
average_vars 3 other papers
2 dated earlier, 1 later
1803.02999 (2018-03)
On First-Order Meta-Learning Algorithms
1904.02957 (2019-04)
Learning to Adapt for Stereo
2010.12777 (2020-10)
Improving Multilingual Models with Language-Clustered Vocabularies
interpolate_vars 3 other papers
2 dated earlier, 1 later
1803.02999 (2018-03)
On First-Order Meta-Learning Algorithms
1904.02957 (2019-04)
Learning to Adapt for Stereo
2010.12777 (2020-10)
Improving Multilingual Models with Language-Clustered Vocabularies
subtract_vars 3 other papers
2 dated earlier, 1 later
1803.02999 (2018-03)
On First-Order Meta-Learning Algorithms
1904.02957 (2019-04)
Learning to Adapt for Stereo
2010.12777 (2020-10)
Improving Multilingual Models with Language-Clustered Vocabularies

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

We build a theoretical framework for designing and understanding practical meta-learning methods that integrates sophisticated formalizations of task-similarity with the extensive literature on online convex optimization and sequential prediction algorithms. Our approach enables the task-similarity to be learned adaptively, provides sharper transfer-risk bounds in the setting of statistical learning-to-learn, and leads to straightforward derivations of average-case regret bounds for efficient algorithms in settings where the task-environment changes dynamically or the tasks share a certain geometric structure. We use our theory to modify several popular meta-learning algorithms and improve their meta-test-time performance on standard problems in few-shot learning and federated learning.

For agents

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

get_harvested_code_for_paper("1906.02717")
get_code_for_paper("1906.02717")
have("1906.02717")

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

curl https://syntology.ai/api/ran/1906.02717.json

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

[![Syntology run record](https://syntology.ai/api/ran/1906.02717.svg)](https://syntology.ai/paper/1906.02717)

Connect an agent — have() is free.