SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.04911 · 2020

GAMA: a General Automated Machine learning Assistant

arXiv · PDF · Open in the Atlas

Code that ran

17 of 17 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 16 have no contract verdict; 1 failure of our instrument; 0 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/2007.04911.json · badge: /api/ran/2007.04911.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
PGijsbers/gama canonical (classifier label)
classifier basis: census
9 ran · 0 no recorded run
jim-schwoebel/allie community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
openml-labs/gama community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
FunctionStatusWhere it lives
classifyfolder Ran
ran on a synthesized input
jim-schwoebel/allie/models/load.py
code served (permissive licence) · get_code("af5e4babbb921931")
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 json
import pickle
import time
import sys
import shutil

# Create a mock listdir with various file types
mock_listdir = [
    'song1.mp3',
    'song2.wav',
    'image1.png',
    'image2.jpg',
    'document.txt',
    'video1.mp4',
    'video2.avi',
    'data.csv',
    'readme.md'
]

ARGS = {
    'listdir': mock_listdir
}
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
evaluate_on_rung Ran
ran on a synthesized input
PGijsbers/gama/gama/search_methods/asha.py
code served (permissive licence) · get_code("57c8a0339bc269d0")
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 MockEvaluation. 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)

class MockIndividual:
    def __init__(self):
        self.meta = {}

class MockEvaluation:
    def __init__(self, individual, error=None):
        self.individual = individual
        self.error = error

def mock_evaluate_individual(individual, *args, **kwargs):
    return MockEvaluation(individual, error=None)

individual = MockIndividual()
rung = 1
max_rung = 3
evaluate_individual = mock_evaluate_individual

ARGS = {
    'individual': individual,
    'rung': rung,
    'max_rung': max_rung,
    'evaluate_individual': evaluate_individual,
    'args': (),
    'kwargs': {'subsample': 0.5}
}
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 MockEvaluation — a MockEvaluation output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MockEvaluation 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
fast_non_dominated_sort Ran
ran on a synthesized input
PGijsbers/gama/gama/genetic_programming/nsga2.py
code served (permissive licence) · get_code("ee60026f2ff14ace")
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

class NSGAMeta:
    """A helper class for comparing data points for NSGA2."""

    def __init__(self, obj: object, metrics):
        self.obj = obj
        self.values = tuple((m(obj) for m in metrics))
        self.rank = 0
        self.distance = 0.0
        self.dominating = []
        self.domination_counter = 0

    def dominates(self, other: 'NSGAMeta') -> bool:
        for self_val, other_val in zip(self.values, other.values):
            if self_val <= other_val:
                return False
        return True

    def crowd_compare(self, other: 'NSGAMeta') -> int:
        """Favor higher rank, if equal, favor less crowded."""
        self_better = self.rank < other.rank or (self.rank == other.rank and self.distance > other.distance)
        return -1 if self_better else 1


torch.manual_seed(0)

# Create simple objects to optimize
obj1 = {"id": 1, "value": 5.0}
obj2 = {"id": 2, "value": 3.0}
obj3 = {"id": 3, "value": 7.0}
obj4 = {"id": 4, "value": 2.0}

# Define metrics (callables that extract values from objects)
def metric1(obj):
    return obj["value"]

def metric2(obj):
    return -obj["value"] * 0.5

metrics = [metric1, metric2]

# Create NSGAMeta instances
nsga_meta_1 = NSGAMeta(obj1, metrics)
nsga_meta_2 = NSGAMeta(obj2, metrics)
nsga_meta_3 = NSGAMeta(obj3, metrics)
nsga_meta_4 = NSGAMeta(obj4, metrics)

# Create population
P = [nsga_meta_1, nsga_meta_2, nsga_meta_3, nsga_meta_4]

ARGS = {
    "P": P
}
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
fast_non_dominated_sort Ran
ran on a synthesized input
openml-labs/gama/gama/genetic_programming/nsga2.py
code served (permissive licence) · get_code("b3f0bed67a0d0b50")
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 itertools
from typing import List, Any, Callable

class NSGAMeta:
    """A helper class for comparing data points for NSGA2."""

    def __init__(self, obj: object, metrics: List[Callable]):
        self.obj = obj
        self.values = tuple((m(obj) for m in metrics))
        self.rank = 0
        self.distance = 0.0
        self.dominating: List['NSGAMeta'] = []
        self.domination_counter = 0

    def dominates(self, other: 'NSGAMeta') -> bool:
        return all((self_val > other_val for self_val, other_val in zip(self.values, other.values)))

    def crowd_compare(self, other: 'NSGAMeta') -> int:
        """Favor higher rank, if equal, favor less crowded."""
        self_better = self.rank < other.rank or (self.rank == other.rank and self.distance > other.distance)
        return -1 if self_better else 1

def fast_non_dominated_sort(P: List[NSGAMeta]) -> List[List[NSGAMeta]]:
    """Sorts P into Pareto fronts."""
    fronts: List[List[NSGAMeta]] = [[]]
    for p, q in itertools.combinations(P, 2):
        if p.dominates(q):
            p.dominating.append(q)
            q.domination_counter += 1
        elif q.dominates(p):
            q.dominating.append(p)
            p.domination_counter += 1
    for p in P:
        if p.domination_counter == 0:
            p.rank = 1
            fronts[0].append(p)
    i = 0
    while fronts[i] != []:
        fronts.append([])
        for p in fronts[i]:
            for q in p.dominating:
                q.domination_counter -= 1
                if q.domination_counter == 0:
                    q.rank = i + 1
                    fronts[i + 1].append(q)
        i += 1
    return fronts

# Create simple objects and metrics for testing
obj1 = {"id": 1}
obj2 = {"id": 2}
obj3 = {"id": 3}
obj4 = {"id": 4}

# Define metrics that return numeric values
metric1 = lambda obj: obj["id"] * 2
metric2 = lambda obj: obj["id"] * 3

# Create NSGAMeta instances
meta1 = NSGAMeta(obj1, [metric1, metric2])
meta2 = NSGAMeta(obj2, [metric1, metric2])
meta3 = NSGAMeta(obj3, [metric1, metric2])
meta4 = NSGAMeta(obj4, [metric1, metric2])

# Create population for fast_non_dominated_sort
P = [meta1, meta2, meta3, meta4]

ARGS = {
    "P": P
}
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 — 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
format_y Ran
ran on a synthesized input
PGijsbers/gama/gama/data_formatting.py
code served (permissive licence) · get_code("0d39f4175f94181d")
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 Series. 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 pandas as pd

# Test case 1: numpy array 1D
y_1d = np.array([0, 1, 0, 1])

# Test case 2: numpy array 2D with multiple columns (will be argmax'd)
y_2d_multi = np.array([[0.1, 0.9], [0.8, 0.2], [0.3, 0.7]])

# Test case 3: numpy array 2D with single column (will be squeezed)
y_2d_single = np.array([[0], [1], [0]])

# Test case 4: pandas Series
y_series = pd.Series([0, 1, 0, 1])

# Test case 5: pandas DataFrame
y_df = pd.DataFrame({'target': [0, 1, 0, 1]})

ARGS = {
    "y": y_1d,
    "y_type": pd.Series
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Series — a Series output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Series output the recorded ndim 1 and dtype family int 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
format_y Ran
ran on a synthesized input
openml-labs/gama/gama/data_formatting.py
code served (permissive licence) · get_code("4666e8ed5c65fb12")
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 Series. 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 pandas as pd

# Test case 1: numpy array 1D
y_1d = np.array([0, 1, 2, 1, 0])

# Test case 2: numpy array 2D with multiple classes
y_2d_multi = np.array([[0.9, 0.1], [0.2, 0.8], [0.3, 0.7]])

# Test case 3: numpy array 2D with single class
y_2d_single = np.array([[1.0], [0.5], [0.8]])

# Test case 4: pandas Series
y_series = pd.Series([0, 1, 2, 1, 0])

# Test case 5: pandas DataFrame
y_dataframe = pd.DataFrame({'target': [0, 1, 2, 1, 0]})

ARGS = {
    'y': y_1d,
    'y_type': pd.Series
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Series — a Series output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Series output the recorded ndim 1 and dtype family int 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_labels Ran
ran on a synthesized input
jim-schwoebel/allie/features/audio_features/yamnet_features.py
code served (permissive licence) · get_code("0659761c261891b8")
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

torch.manual_seed(0)

ARGS = {
    'vector': [1, 2, 3, 4, 5],
    'label': 'sample_',
    'label2': 'type_A'
}
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
load_csv_header Ran
ran on a synthesized input
PGijsbers/gama/gama/data_loading.py
code served (permissive licence) · get_code("520cd8123a24aacd")
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 tempfile
import os

# Create a temporary CSV file with a header
temp_dir = tempfile.mkdtemp()
csv_file_path = os.path.join(temp_dir, "test_data.csv")

with open(csv_file_path, 'w') as f:
    f.write("name,age,city\n")
    f.write("Alice,30,NYC\n")
    f.write("Bob,25,LA\n")

ARGS = {
    'file_path': csv_file_path
}
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
most_common Ran
ran; our contract draft was wrong, not the code
jim-schwoebel/allie/models/load.py
code served (permissive licence) · get_code("1deecc25893dafad")
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)

# Analyze the function:
# - most_common(lst) takes a list parameter
# - It uses set(lst) to get unique items
# - It uses lst.count() method on items in the set
# - It returns the item with the maximum count
# - The list should contain hashable items that can be counted

# Create a simple list with repeated items
lst = [1, 2, 2, 3, 2, 4, 1, 2, 5]

ARGS = {
    "lst": lst
}
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
nsga2 Ran
ran on a synthesized input
PGijsbers/gama/gama/genetic_programming/nsga2.py
code served (permissive licence) · get_code("a040787fe7e3a043")
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 numpy as np

# Create a simple population of objects (dicts with some properties)
population = [
    {"accuracy": 0.85, "loss": 0.15, "speed": 0.9},
    {"accuracy": 0.82, "loss": 0.18, "speed": 0.95},
    {"accuracy": 0.88, "loss": 0.12, "speed": 0.85},
    {"accuracy": 0.80, "loss": 0.20, "speed": 1.0},
    {"accuracy": 0.86, "loss": 0.14, "speed": 0.88},
    {"accuracy": 0.84, "loss": 0.16, "speed": 0.92},
]

# Define metrics (callables that extract float values from population objects)
# We want to maximize accuracy and speed, minimize loss
metrics = [
    lambda obj: obj["accuracy"],      # higher is better
    lambda obj: obj["speed"],         # higher is better
    lambda obj: -obj["loss"],         # lower loss is better, so negate
]

n = 3
return_meta = False

ARGS = {
    "population": population,
    "n": n,
    "metrics": metrics,
    "return_meta": return_meta,
}
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 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
nsga2 Ran
ran on a synthesized input
openml-labs/gama/gama/genetic_programming/nsga2.py
code served (permissive licence) · get_code("8e229aa4f8a8f56d")
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 numpy as np

# Create a simple population of objects with numerical properties
class Individual:
    def __init__(self, x, y, z):
        self.x = x
        self.y = y
        self.z = z

# Create a small population
population = [
    Individual(1.0, 2.0, 3.0),
    Individual(2.0, 1.5, 2.5),
    Individual(1.5, 3.0, 2.0),
    Individual(3.0, 1.0, 1.5),
    Individual(2.5, 2.5, 2.5),
    Individual(0.5, 3.5, 3.5),
]

# Define metrics as callables that extract numeric values from individuals
metrics = [
    lambda ind: ind.x,
    lambda ind: ind.y,
    lambda ind: -ind.z,  # Negate z to demonstrate multi-objective optimization
]

# Number of individuals to select
n = 3

# Return meta information or not
return_meta = False

ARGS = {
    "population": population,
    "n": n,
    "metrics": metrics,
    "return_meta": return_meta,
}
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 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
nsga2_select Ran
ran on a synthesized input
PGijsbers/gama/gama/genetic_programming/nsga2.py
code served (permissive licence) · get_code("a18c1c1f6a83f5ae")
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 random
import numpy as np

# Set seed for reproducibility
random.seed(42)
np.random.seed(42)

# Create a simple population of objects
# Each object will be a dict with some properties
class SimpleIndividual:
    def __init__(self, x, y):
        self.x = x
        self.y = y

# Create a population of 10 individuals
population = [SimpleIndividual(x=random.uniform(0, 10), y=random.uniform(0, 10)) for _ in range(10)]

# Define metrics as callables that extract values from individuals
# We want to minimize both x and y (so we negate them to turn minimization into maximization for NSGA2)
metric1 = lambda ind: -ind.x  # Negative because we want to maximize (for domination)
metric2 = lambda ind: -ind.y

metrics = [metric1, metric2]

# Number of pairs to select
n = 4

ARGS = {
    'population': population,
    'n': n,
    'metrics': metrics
}
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
nsga2_select Ran
ran on a synthesized input
openml-labs/gama/gama/genetic_programming/nsga2.py
code served (permissive licence) · get_code("a7f96752452ada1c")
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 3, 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 random
import numpy as np

def metric_1(obj):
    """First metric: maximize value at index 0"""
    return obj[0]

def metric_2(obj):
    """Second metric: maximize value at index 1"""
    return obj[1]

def metric_3(obj):
    """Third metric: maximize value at index 2"""
    return obj[2]

# Create a population of simple numeric tuples
population = [
    (1.0, 2.0, 3.0),
    (2.0, 1.5, 2.5),
    (1.5, 3.0, 1.0),
    (3.0, 2.5, 2.0),
    (2.5, 2.5, 2.5),
    (1.0, 1.0, 1.0),
]

# Define metrics as callables
metrics = [metric_1, metric_2, metric_3]

# Number of pairs to select
n = 2

ARGS = {
    'population': population,
    'n': n,
    'metrics': metrics,
}
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 3, 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
numpy_to_dataframe Ran
ran on a synthesized input
PGijsbers/gama/gama/data_formatting.py
code served (permissive licence) · get_code("c2b84360c1f09a1d")
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 DataFrame. 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 pandas as pd

# Create a simple numpy array to pass to numpy_to_dataframe
# Mix of integer and float values to test the categorical inference
x = np.array([
    [1, 2, 5, 10],
    [1, 3, 5, 20],
    [2, 2, 6, 10],
    [3, 4, 7, 30]
], dtype=np.int64)

ARGS = {
    'x': x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family int 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
prev_dir Ran
ran on a synthesized input
jim-schwoebel/allie/models/load.py
code served (permissive licence) · get_code("8e13be73f784da6b")
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 str. 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

def prev_dir(directory):
    g = directory.split('/')
    dir_ = ''
    for i in range(len(g)):
        if i != len(g) - 1:
            if i == 0:
                dir_ = dir_ + g[i]
            else:
                dir_ = dir_ + '/' + g[i]
    return dir_

ARGS = {
    'directory': '/home/user/documents/file.txt'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
series_looks_categorical Ran
ran on a synthesized input
PGijsbers/gama/gama/data_formatting.py
code served (permissive licence) · get_code("8424920875ff39be")
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 pandas as pd
import numpy as np
from pandas.core.dtypes.common import is_numeric_dtype

# Create test series for the function
# Test case 1: object dtype (should return True)
series_object = pd.Series(['a', 'b', 'c', 'a', 'b'])

# Test case 2: integer dtype with <= 10 unique values (should return True)
series_int_categorical = pd.Series([1, 2, 3, 1, 2, 3, 1, 2], dtype='int64')

# Test case 3: float dtype with integer-like values and <= 10 unique values (should return True)
series_float_categorical = pd.Series([1.0, 2.0, 3.0, 1.0, 2.0, 3.0], dtype='float64')

# Test case 4: numeric dtype with > 10 unique values (should return False)
series_numeric_many = pd.Series(np.arange(20), dtype='int64')

# Test case 5: float dtype with non-integer values (should return False)
series_float_continuous = pd.Series([1.5, 2.7, 3.2, 4.1], dtype='float64')

# Use the first one as the main test case
ARGS = {
    'series': series_int_categorical
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
sniff_csv_meta Ran
ran on a synthesized input
PGijsbers/gama/gama/data_loading.py
code served (permissive licence) · get_code("cf3c99ff20bab310")
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 csv
import tempfile
import os

# Create a temporary CSV file with valid content
temp_dir = tempfile.mkdtemp()
temp_csv_path = os.path.join(temp_dir, "test.csv")

# Write a sample CSV file with header and comma delimiter
with open(temp_csv_path, 'w', newline='') as f:
    writer = csv.writer(f, delimiter=',')
    writer.writerow(['name', 'age', 'city'])
    writer.writerow(['Alice', '30', 'New York'])
    writer.writerow(['Bob', '25', 'Los Angeles'])
    writer.writerow(['Charlie', '35', 'Chicago'])

ARGS = {
    'file_path': temp_csv_path
}
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, 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

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 13 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
most_common 3 other papers
1 dated earlier, 2 later
1803.08823 (2018-03)
A high-bias, low-variance introduction to Machine Learning for physicists
2106.09248 (2021-06)
X-FACT: A New Benchmark Dataset for Multilingual Fact Checking
2209.02071 (2022-09)
CONCRETE: Improving Cross-lingual Fact-checking with Cross-lingual Retrieval

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

The General Automated Machine learning Assistant (GAMA) is a modular AutoML system developed to empower users to track and control how AutoML algorithms search for optimal machine learning pipelines, and facilitate AutoML research itself. In contrast to current, often black-box systems, GAMA allows users to plug in different AutoML and post-processing techniques, logs and visualizes the search process, and supports easy benchmarking. It currently features three AutoML search algorithms, two model post-processing steps, and is designed to allow for more components to be added.

For agents

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

get_harvested_code_for_paper("2007.04911")
get_code_for_paper("2007.04911")
have("2007.04911")

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

curl https://syntology.ai/api/ran/2007.04911.json

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

[![Syntology run record](https://syntology.ai/api/ran/2007.04911.svg)](https://syntology.ai/paper/2007.04911)

Connect an agent — have() is free.