SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.05826 · 2022

End-to-end Algorithm Synthesis with Recurrent Networks: Logical Extrapolation Without Overthinking

arXiv · PDF · Open in the Atlas

Code that ran

3 of 18 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 0 failures of our instrument; 15 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/2202.05826.json · badge: /api/ran/2202.05826.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
aks2203/deep-thinking canonical (classifier label)
classifier basis not recorded on the link
3 ran · 15 no recorded run
FunctionStatusWhere it lives
get_little_df_from_one_run Ran
ran on a synthesized input
aks2203/deep-thinking/deepthinking/data_analysis/make_table.py
code served (permissive licence) · get_code("077d9b81c2525794")
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 copy
import pandas as pd

# Construct a valid data_dict argument
data_dict = {
    'test_iters': [100, 200, 300],
    'test_acc': {
        '100': 0.85,
        '200': 0.88,
        '300': 0.90
    },
    'val_acc': {
        '100': 0.80,
        '200': 0.83,
        '300': 0.86
    },
    'train_acc': {
        '100': 0.92,
        '200': 0.94,
        '300': 0.95
    },
    'model_name': 'test_model',
    'learning_rate': 0.001,
    'batch_size': 32
}

ARGS = {
    'data_dict': data_dict
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
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 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_schoopy_plot Ran
ran on a synthesized input
aks2203/deep-thinking/deepthinking/data_analysis/make_schoop.py
code served (permissive licence) · get_code("5b24b1cc9133ef6d")
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 Axes. 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 pandas as pd
import numpy as np

# Create a small dataframe that matches what get_schoopy_plot expects
np.random.seed(0)

# Generate sample data with the required columns
n_rows = 20
table_data = {
    'test_iter': np.linspace(0, 100, n_rows),
    'test_acc_mean': np.random.rand(n_rows) * 100,
    'test_acc_sem': np.random.rand(n_rows) * 2,
    'model': np.random.choice(['model_a', 'model_b'], n_rows),
    'test_data': np.random.choice(['dataset_1', 'dataset_2'], n_rows),
    'alpha': np.random.choice([0.001, 0.01], n_rows),
    'max_iters': np.full(n_rows, 100)
}

table = pd.DataFrame(table_data)

ARGS = {
    'table': table,
    'error_bars': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, pandas, seaborn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Axes — a Axes output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Axes 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_trained_checkpoints Ran
ran on a synthesized input
aks2203/deep-thinking/deepthinking/data_analysis/parse_training_runs.py
code served (permissive licence) · get_code("96cfe7a14410ae6d")
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 tempfile
import json
import os
from pathlib import Path
from omegaconf import OmegaConf

# Create a temporary directory structure with mock training data
tmpdir = tempfile.mkdtemp()

# Create directory structure: filepath/model_name/training_1/.hydra/config.yaml and stats.json
model_dir = os.path.join(tmpdir, "model_exp", "training_1")
hydra_dir = os.path.join(model_dir, ".hydra")
os.makedirs(hydra_dir, exist_ok=True)

# Create a mock config.yaml
config_dict = {
    "problem": {
        "model": {
            "model": "resnet50"
        },
        "hyp": {
            "alpha": 0.1
        }
    }
}
config_yaml_path = os.path.join(hydra_dir, "config.yaml")
OmegaConf.save(OmegaConf.create(config_dict), config_yaml_path)

# Create a mock stats.json
stats_data = {
    "max_iters": 100,
    "train_acc": {
        "100": 0.95
    }
}
stats_json_path = os.path.join(model_dir, "stats.json")
with open(stats_json_path, 'w') as f:
    json.dump(stats_data, f)

# Create another checkpoint with different model/alpha
model_dir2 = os.path.join(tmpdir, "model_exp2", "training_2")
hydra_dir2 = os.path.join(model_dir2, ".hydra")
os.makedirs(hydra_dir2, exist_ok=True)

config_dict2 = {
    "problem": {
        "model": {
            "model": "vgg16"
        },
        "hyp": {
            "alpha": 0.2
        }
    }
}
config_yaml_path2 = os.path.join(hydra_dir2, "config.yaml")
OmegaConf.save(OmegaConf.create(config_dict2), config_yaml_path2)

stats_data2 = {
    "max_iters": 100,
    "train_acc": {
        "100": 0.92
    }
}
stats_json_path2 = os.path.join(model_dir2, "stats.json")
with open(stats_json_path2, 'w') as f:
    json.dump(stats_data2, f)

ARGS = {
    "filepath": tmpdir,
    "acc_filter": 0.85,
    "model_list": ["resnet50", "vgg16"],
    "alpha_list": [0.1, 0.2]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
omegaconf — 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
dt_net_1d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/dt_net_1d.py
code served (permissive licence) · get_code("f7945168c1ba42ae")
How it ran

no invocation recorded.

dt_net_2d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/dt_net_2d.py
code served (permissive licence) · get_code("72bc1b13ede1f48f")
How it ran

no invocation recorded.

dt_net_gn_1d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/dt_net_1d.py
code served (permissive licence) · get_code("cb8d7cd07dcb373e")
How it ran

no invocation recorded.

dt_net_gn_2d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/dt_net_2d.py
code served (permissive licence) · get_code("3157b448da0db767")
How it ran

no invocation recorded.

dt_net_recall_1d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/dt_net_1d.py
code served (permissive licence) · get_code("2745fbdd05dd78c5")
How it ran

no invocation recorded.

dt_net_recall_2d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/dt_net_2d.py
code served (permissive licence) · get_code("5efe7003a1ae9fce")
How it ran

no invocation recorded.

feedforward_net_1d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/feedforward_net_1d.py
code served (permissive licence) · get_code("d906e24e377f708e")
How it ran

no invocation recorded.

feedforward_net_2d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/feedforward_net_2d.py
code served (permissive licence) · get_code("e88c13ba29fae08d")
How it ran

no invocation recorded.

feedforward_net_gn_1d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/feedforward_net_1d.py
code served (permissive licence) · get_code("fffa340d37ae79ea")
How it ran

no invocation recorded.

feedforward_net_gn_2d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/feedforward_net_2d.py
code served (permissive licence) · get_code("0359c5b8a9eddda2")
How it ran

no invocation recorded.

feedforward_net_recall_1d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/feedforward_net_1d.py
code served (permissive licence) · get_code("c31bd413c9229009")
How it ran

no invocation recorded.

feedforward_net_recall_2d Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/models/feedforward_net_2d.py
code served (permissive licence) · get_code("a0cc6c49618902d6")
How it ran

no invocation recorded.

get_df Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/data_analysis/make_table.py
code served (permissive licence) · get_code("04c48fd8b7c02d03")
How it ran

no invocation recorded.

get_table Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/data_analysis/make_table.py
code served (permissive licence) · get_code("340e568d6e6480c5")
How it ran

no invocation recorded.

get_trained_checkpoints Not yet run
unverified (harvested, no recorded run)
aks2203/deep-thinking/deepthinking/data_analysis/get_names_and_info.py
code served (permissive licence) · get_code("2a9ba28a21829bcb")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 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 18 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

Machine learning systems perform well on pattern matching tasks, but their ability to perform algorithmic or logical reasoning is not well understood. One important reasoning capability is algorithmic extrapolation, in which models trained only on small/simple reasoning problems can synthesize complex strategies for large/complex problems at test time. Algorithmic extrapolation can be achieved through recurrent systems, which can be iterated many times to solve difficult reasoning problems. We observe that this approach fails to scale to highly complex problems because behavior degenerates when many iterations are applied -- an issue we refer to as "overthinking." We propose a recall architecture that keeps an explicit copy of the problem instance in memory so that it cannot be forgotten. We also employ a progressive training routine that prevents the model from learning behaviors that are specific to iteration number and instead pushes it to learn behaviors that can be repeated indefinitely. These innovations prevent the overthinking problem, and enable recurrent systems to solve extremely hard extrapolation tasks.

For agents

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

get_harvested_code_for_paper("2202.05826")
get_code_for_paper("2202.05826")
have("2202.05826")

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

curl https://syntology.ai/api/ran/2202.05826.json

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

[![Syntology run record](https://syntology.ai/api/ran/2202.05826.svg)](https://syntology.ai/paper/2202.05826)

Connect an agent — have() is free.