SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1903.03096 · 2019

Meta-Dataset: A Dataset of Datasets for Learning to Learn from Few Examples

arXiv · PDF · Open in the Atlas

Code that ran

10 of 11 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; 1 honoured the contract we drafted, 0 violated it and 6 have no contract verdict; 3 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/1903.03096.json · badge: /api/ran/1903.03096.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/meta-dataset canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
ebadrian/metadl community (classifier label)
classifier basis not recorded on the link
6 ran · 1 no recorded run
nobody-1617/deta extension (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
compute_class_precision Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
google-research/meta-dataset/meta_dataset/analyze.py
code served (permissive licence) · get_code("6b4ce423f84ce9bc")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned float64, 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 numpy as np

np.random.seed(0)

num_test_examples = 10
num_classes = 5
class_id = 2

logits = np.random.randn(num_test_examples, num_classes).astype(np.float32)
targets = np.random.randint(0, num_classes, size=(num_test_examples, num_classes))

ARGS = {
    'class_id': class_id,
    'logits': logits,
    'targets': targets
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, 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
compute_episode_accuracy Ran
ran, honoured the contract we drafted
google-research/meta-dataset/meta_dataset/analyze.py
code served (permissive licence) · get_code("6a9728ff66fde324")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned float64, 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 numpy as np

# Set up test data with distinct dimensions
num_test_examples = 5
num_classes = 3

# Create logits: shape [num_test_examples, num_classes]
logits = np.array([
    [2.1, 0.5, 0.3],
    [0.2, 3.0, 0.1],
    [0.1, 0.2, 2.5],
    [1.8, 0.4, 0.6],
    [0.3, 2.2, 0.5]
], dtype=np.float32)

# Create targets: shape [num_test_examples] with class indices
targets = np.array([0, 1, 2, 0, 1], dtype=np.int32)

ARGS = {
    'logits': logits,
    'targets': targets
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, 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
conv_net Ran
ran on a synthesized input
ebadrian/metadl/baselines/maml/helper.py
code served (permissive licence) · get_code("ab96a11f341b7378")
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 Sequential. 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 tensorflow as tf

ARGS = {
    "nbr_classes": 5,
    "img_size": 28,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential 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
extract_elapsed_time Ran
ran on a synthesized input
ebadrian/metadl/metadl/core/scoring/scoring.py
code served (permissive licence) · get_code("568b7bd83c95b525")
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 os
import tempfile

# Create a temporary directory with a metadata file
temp_dir = tempfile.mkdtemp()
metadata_path = os.path.join(temp_dir, 'metadata')

# Write a metadata file with elapsed time information
with open(metadata_path, 'w') as f:
    f.write('some_key value another_key value elapsed 42.5 other_stuff\n')

ARGS = {
    'saved_model_dir': temp_dir
}
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 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
get_logger Ran
ran on a synthesized input
ebadrian/metadl/metadl/core/parent_scoring/parent_scoring.py
code served (permissive licence) · get_code("e8e595468bf6e7e6")
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 Logger. 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 sys
import logging

ARGS = {
    'verbosity_level': 'INFO',
    'use_error_log': False
}
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 Logger — a Logger output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Logger 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
get_shot_to_precision Ran
ran; our contract draft was wrong, not the code
google-research/meta-dataset/meta_dataset/analyze.py
code served (permissive licence) · get_code("b2efd9150822b76f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned defaultdict. 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 numpy as np
import collections

# Set up a simple few-shot learning scenario with 2 episodes
# Episode 1: 3-way classification with varying shots
# Episode 2: 4-way classification with varying shots

# Episode 1: 3 classes, test set size = 6
episode1_shots = np.array([1, 2, 3])  # class 0 has 1 shot, class 1 has 2 shots, class 2 has 3 shots
episode1_logits = np.array([
    [0.8, 0.1, 0.1],  # predicts class 0
    [0.1, 0.7, 0.2],  # predicts class 1
    [0.2, 0.1, 0.7],  # predicts class 2
    [0.6, 0.3, 0.1],  # predicts class 0
    [0.1, 0.8, 0.1],  # predicts class 1
    [0.1, 0.2, 0.7],  # predicts class 2
])
episode1_targets = np.array([0, 1, 2, 0, 1, 2])

# Episode 2: 4 classes, test set size = 8
episode2_shots = np.array([2, 2, 3, 1])  # class 0 has 2 shots, class 1 has 2 shots, class 2 has 3 shots, class 3 has 1 shot
episode2_logits = np.array([
    [0.9, 0.05, 0.03, 0.02],  # predicts class 0
    [0.1, 0.8, 0.05, 0.05],   # predicts class 1
    [0.05, 0.1, 0.8, 0.05],   # predicts class 2
    [0.05, 0.05, 0.1, 0.8],   # predicts class 3
    [0.7, 0.1, 0.1, 0.1],     # predicts class 0
    [0.1, 0.7, 0.1, 0.1],     # predicts class 1
    [0.1, 0.1, 0.7, 0.1],     # predicts class 2
    [0.1, 0.1, 0.1, 0.7],     # predicts class 3
])
episode2_targets = np.array([0, 1, 2, 3, 0, 1, 2, 3])

shots = [episode1_shots, episode2_shots]
logits = [episode1_logits, episode2_logits]
targets = [episode1_targets, episode2_targets]

ARGS = {
    'shots': shots,
    'logits': logits,
    'targets': targets
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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
is_one_hot_vector Ran
ran on a synthesized input
ebadrian/metadl/metadl/core/scoring/scoring.py
code served (permissive licence) · get_code("5c8ffb087acc6095")
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 bool, 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 numpy as np

# Create test cases for is_one_hot_vector
# The function checks if vectors are one-hot encoded (one entry is 1, rest are 0)

# Case 1: Single one-hot vector
one_hot_vector = np.array([1, 0, 0, 0])

# Case 2: Batch of vectors (some one-hot, some not)
batch_vectors = np.array([
    [1, 0, 0, 0],
    [0, 1, 0, 0],
    [0, 0, 1, 0],
    [0, 0, 0, 1]
])

# Case 3: Matrix with axis parameter
matrix_2d = np.array([
    [1, 0, 0],
    [0, 1, 0],
    [0, 0, 1]
])

# Use the simplest case: a single one-hot vector with default parameters
ARGS = {
    "x": one_hot_vector,
    "axis": None,
    "keepdims": False
}
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 bool, 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
moving_average Ran
ran; our contract draft was wrong, not the code
nobody-1617/deta/meta_dataset/analysis/select_best_model.py
code served (permissive licence) · get_code("9ff47b55c732dbfd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned ndarray, ndim 1, dtype family float, shape (10,) 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

# Create a simple 1D array to smooth
x = np.array([1.0, 2.0, 5.0, 3.0, 4.0, 6.0, 2.0, 1.0, 3.0, 5.0])

# Define a smooth window size
smooth_window = 3

ARGS = {
    'x': x,
    'smooth_window': smooth_window
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
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 1, dtype family float, shape (10,) — shape (10,) 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
read_curve Ran
ran on a synthesized input
ebadrian/metadl/metadl/core/parent_scoring/parent_scoring.py
code served (permissive licence) · get_code("eb381c3ecf39155c")
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 os
from os.path import join
from glob import glob
import tempfile
import shutil

# Create a temporary directory structure with mock learning curve files
temp_dir = tempfile.mkdtemp()

# Create the expected directory structure
for i in range(5):
    res_dir = join(temp_dir, 'res_' + str(i + 2))
    os.makedirs(res_dir, exist_ok=True)
    # Create a dummy learning curve PNG file
    dummy_file = join(res_dir, 'learning-curve-001.png')
    with open(dummy_file, 'w') as f:
        f.write('dummy')

# Create a simple args class
class Args:
    def __init__(self, input_dir):
        self.input_dir = input_dir

ARGS = {
    'args': Args(input_dir=temp_dir)
}
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 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
read_score Ran
ran on a synthesized input
ebadrian/metadl/metadl/core/parent_scoring/parent_scoring.py
code served (permissive licence) · get_code("2f30e207b048cceb")
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 os
import tempfile
import yaml
from pathlib import Path

# Create a temporary directory structure with score files
temp_dir = tempfile.mkdtemp()

# Create 5 result directories with scores.txt files
for i in range(5):
    res_dir = os.path.join(temp_dir, f'res_{i + 2}')
    os.makedirs(res_dir, exist_ok=True)
    
    # Write a scores.txt file with yaml format
    scores_file = os.path.join(res_dir, 'scores.txt')
    score_data = {'score': 0.75 + i * 0.05}
    with open(scores_file, 'w') as f:
        yaml.dump(score_data, f)

# Create a simple args object
class Args:
    def __init__(self, input_dir):
        self.input_dir = input_dir

ARGS = {
    'args': Args(input_dir=temp_dir)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
yaml — 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
NwayKshot_accuracy Not yet run
unverified (harvested, no recorded run)
ebadrian/metadl/metadl/core/scoring/scoring.py
code served (permissive licence) · get_code("cd5ee3f06fa23554")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 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 11 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

Few-shot classification refers to learning a classifier for new classes given only a few examples. While a plethora of models have emerged to tackle it, we find the procedure and datasets that are used to assess their progress lacking. To address this limitation, we propose Meta-Dataset: a new benchmark for training and evaluating models that is large-scale, consists of diverse datasets, and presents more realistic tasks. We experiment with popular baselines and meta-learners on Meta-Dataset, along with a competitive method that we propose. We analyze performance as a function of various characteristics of test tasks and examine the models' ability to leverage diverse training sources for improving their generalization. We also propose a new set of baselines for quantifying the benefit of meta-learning in Meta-Dataset. Our extensive experimentation has uncovered important research challenges and we hope to inspire work in these directions.

For agents

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

get_harvested_code_for_paper("1903.03096")
get_code_for_paper("1903.03096")
have("1903.03096")

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

curl https://syntology.ai/api/ran/1903.03096.json

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

[![Syntology run record](https://syntology.ai/api/ran/1903.03096.svg)](https://syntology.ai/paper/1903.03096)

Connect an agent — have() is free.