SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.20082 · ICLR · 2024

Efficient Subgraph GNNs by Learning Effective Selection Policies

Haggai Maron, Bruno Ribeiro, Eli Meirom, Beatrice Bevilacqua, Moshe Eliasof

arXiv · PDF · Open in the Atlas

Code that ran

7 of 9 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 0 have no contract verdict; 7 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/2310.20082.json · badge: /api/ran/2310.20082.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
beabevi/policy-learn canonical (classifier label)
classifier basis: census
7 ran · 2 no recorded run
FunctionStatusWhere it lives
lengths2slices Ran
ran; our contract draft was wrong, not the code
beabevi/policy-learn/data_utils.py
code served (permissive licence) · get_code("c6920406a0f53f26")
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 Tensor, ndim 1, dtype family int, shape (5,) 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

torch.manual_seed(0)

# lengths2slices expects a LongTensor of lengths
# The function calls .cumsum(-1) on it and stacks with a 0 tensor
# So we need a 1D or multi-dimensional LongTensor

lengths = torch.tensor([2, 3, 1, 4], dtype=torch.int64)

ARGS = {
    "lengths": lengths
}
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 Tensor, ndim 1, dtype family int, shape (5,) — shape (5,) 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
observation_coos Ran
ran; our contract draft was wrong, not the code
beabevi/policy-learn/observation.py
code served (permissive licence) · get_code("2e56a3afa91eb04e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from typing import TypedDict

torch.manual_seed(0)

class Observation(TypedDict):
    graph_id: torch.Tensor
    which_subgraphs: torch.Tensor
    ys: torch.Tensor

# Create small tensors with distinct dimensions
# graph_id: [task_bsz=2, 1]
graph_id = torch.tensor([[0], [1]], dtype=torch.long)

# which_subgraphs: [task_bsz=2, max_subgraphs=4]
# Use -1 as padding value, and valid subgraph indices otherwise
which_subgraphs = torch.tensor([
    [0, 1, 2, -1],
    [0, 1, -1, -1]
], dtype=torch.long)

# ys: [task_bsz=2, y_d=3]
ys = torch.randn(2, 3)

obs = Observation(
    graph_id=graph_id,
    which_subgraphs=which_subgraphs,
    ys=ys
)

ARGS = {
    'obs': obs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
observation_to Ran
ran; our contract draft was wrong, not the code
beabevi/policy-learn/observation.py
code served (permissive licence) · get_code("e5f2de0d3aba87a7")
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
from typing import TypedDict

torch.manual_seed(0)

class Tensor2D(torch.Tensor):
    pass

class Observation(TypedDict):
    graph_id: torch.Tensor
    which_subgraphs: torch.Tensor
    ys: torch.Tensor

# Create sample observation with distinct dimensions
obs = Observation(
    graph_id=torch.randint(0, 10, (2, 1)),  # task_bsz=2, 1
    which_subgraphs=torch.randint(0, 5, (2, 3)),  # task_bsz=2, max_subgraphs=3
    ys=torch.randn(2, 4)  # task_bsz=2, y_d=4
)

device = torch.device('cpu')

ARGS = {
    'obs': obs,
    'device': device
}
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
resolve Ran
ran; our contract draft was wrong, not the code
beabevi/policy-learn/config.py
code served (permissive licence) · get_code("643fd0cde93c0eaf")
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 DictConfig. 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 omegaconf
from omegaconf import MISSING, OmegaConf

# Create a simple DictConfig with some nested structure and interpolations
base_config = OmegaConf.create({
    "model": {
        "name": "test_model",
        "layers": 3,
        "hidden_size": "${model.layers}"
    },
    "training": {
        "batch_size": 32,
        "epochs": 10,
        "learning_rate": 0.001
    },
    "path": "${model.name}_checkpoint"
})

ARGS = {
    "config": base_config
}
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 DictConfig — a DictConfig output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DictConfig output the recorded ndim 1 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
root_mean_squared_error Ran
ran; our contract draft was wrong, not the code
beabevi/policy-learn/utils.py
code served (permissive licence) · get_code("b2ea0b67cf7e7952")
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 Tensor, 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 torch
import torchmetrics

torch.manual_seed(0)

# Create tiny prediction and target tensors
# batch_size=2, num_features=4
preds = torch.randn(2, 4)
target = torch.randn(2, 4)

ARGS = {
    "preds": preds,
    "target": target,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchmetrics — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, 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
top_k_gumbel_softmax Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
beabevi/policy-learn/utils.py
code served (permissive licence) · get_code("fecd371d49fabda8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a small logits tensor: batch_size=2, num_classes=4
logits = torch.randn(2, 4)

ARGS = {
    "logits": logits,
    "k": 2,
    "tau": 1.0,
    "hard": True,
    "eps": 1e-10,
    "dim": -1,
    "use_noise": True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
torch_vrange Ran
ran; our contract draft was wrong, not the code
beabevi/policy-learn/extensions.py
code served (permissive licence) · get_code("edab9c86df31b750")
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 Tensor, ndim 1, dtype family int, shape (5,) 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

torch.manual_seed(0)

# Construct arguments for torch_vrange
# The function takes lengths and optional starts
# lengths: torch.array of lengths for each range
# starts: torch.array of starts for each range (optional, defaults to None)

# Create a small example with 4 ranges
lengths = torch.tensor([0, 2, 3, 0], dtype=torch.long)
starts = torch.tensor([1, 3, 4, 6], dtype=torch.long)

ARGS = {
    'lengths': lengths,
    'starts': starts,
}
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 Tensor, ndim 1, dtype family int, shape (5,) — shape (5,) 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
fqn_import Not yet run
unverified (harvested, no recorded run)
beabevi/policy-learn/config.py
code served (permissive licence) · get_code("6823f6e0df096019")
How it ran

no invocation recorded.

setup Not yet run
unverified (harvested, no recorded run)
beabevi/policy-learn/utils.py
code served (permissive licence) · get_code("08e7484b677b8f03")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 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 9 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

Subgraph GNNs are provably expressive neural architectures that learn graph representations from sets of subgraphs. Unfortunately, their applicability is hampered by the computational complexity associated with performing message passing on many subgraphs. In this paper, we consider the problem of learning to select a small subset of the large set of possible subgraphs in a data-driven fashion. We first motivate the problem by proving that there are families of WL-indistinguishable graphs for which there exist efficient subgraph selection policies: small subsets of subgraphs that can already identify all the graphs within the family. We then propose a new approach, called POLICY-LEARN, that learns how to select subgraphs in an iterative manner. We prove that, unlike popular random policies and prior work addressing the same problem, our architecture is able to learn the efficient policies mentioned above. Our experimental results demonstrate that POLICY-LEARN outperforms existing baselines across a wide range of datasets.

For agents

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

get_harvested_code_for_paper("2310.20082")
get_code_for_paper("2310.20082")
have("2310.20082")

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

curl https://syntology.ai/api/ran/2310.20082.json

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

[![Syntology run record](https://syntology.ai/api/ran/2310.20082.svg)](https://syntology.ai/paper/2310.20082)

Connect an agent — have() is free.