SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.04910 · NeurIPS · 2022

Instance-wise algorithm configuration with graph neural networks

Max Paulus, Romeo Valentin, Claudio Ferrari, Jérémy Scheurer, Andisheh Amrollahi, Chris Wendler

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 0 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/2202.04910.json · badge: /api/ran/2202.04910.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
romeov/ml4co-competition canonical (classifier label)
classifier basis: classifier_2025
4 ran · 1 no recorded run
FunctionStatusWhere it lives
ConfigEmbedding Ran
ran on a synthesized input · constructed an object; no result was computed
romeov/ml4co-competition/bo_gnn/config/models/baseline.py
code served (permissive licence) · get_code("6e7c5d3840b9eb94")
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 an instance of its own class, ConfigEmbedding: the run constructed the object and did not call it, so no result was computed. 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

torch.manual_seed(0)

ARGS = {
    'in_dim': 4,
    'hidden_dim': 16,
    'out_dim': 8,
}
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 ConfigEmbedding — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ConfigEmbedding 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
GNNFwd Ran
ran on a synthesized input · constructed an object; no result was computed
romeov/ml4co-competition/bo_gnn/config/models/baseline.py
code served (permissive licence) · get_code("cfa103c5e498589a")
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 an instance of its own class, GNNFwd: the run constructed the object and did not call it, so no result was computed. 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_geometric as tg
from torch_geometric.data import Data

torch.manual_seed(0)

# Create a minimal data object that the forward method expects
num_nodes = 3
num_constraints = 2
num_edges = 4

# Node features: shape (num_nodes, in_dim[0])
var_feats = torch.randn(num_nodes, 4)

# Constraint features: shape (num_constraints, in_dim[1])
cstr_feats = torch.randn(num_constraints, 5)

# Edge index: bipartite graph between nodes and constraints
edge_index = torch.tensor([
    [0, 1, 2, 0],  # node indices
    [0, 1, 0, 1]   # constraint indices
], dtype=torch.long)

# Edge attributes: shape (num_edges,)
edge_attr = torch.randn(num_edges)

# Create a Data object with all required attributes
data = Data(
    var_feats=var_feats,
    cstr_feats=cstr_feats,
    edge_index=edge_index,
    edge_attr=edge_attr
)

ARGS = {
    'in_dim': (4, 5),
    'out_dim': (4, 5),
    'residual': True,
    'batch_norm': True,
    'additional_dense': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GNNFwd — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GNNFwd 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
MilpGNN Ran
ran on a synthesized input · constructed an object; no result was computed
romeov/ml4co-competition/bo_gnn/config/models/baseline.py
code served (permissive licence) · get_code("e5024b69ac719f7f")
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 an instance of its own class, MilpGNN: the run constructed the object and did not call it, so no result was computed. 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_geometric as tg
from torch_geometric.data import Data

torch.manual_seed(0)

# Create a minimal data object that satisfies the forward pass requirements
def create_minimal_data():
    # Number of variables and constraints
    n_vars = 5
    n_cstrs = 3
    batch_size = 2
    
    # Variable features: 9-dimensional (as specified in first GNNFwd layer)
    var_feats = torch.randn(n_vars, 9)
    
    # Constraint features: 1-dimensional (as specified in first GNNFwd layer)
    cstr_feats = torch.randn(n_cstrs, 1)
    
    # Edge index connecting variables and constraints
    # Create a simple bipartite graph
    edge_index = torch.tensor([
        [0, 1, 2, 0, 1],  # constraint indices
        [0, 1, 2, 3, 4]   # variable indices
    ], dtype=torch.long)
    
    # Edge attributes (weights)
    edge_attr = torch.randn(edge_index.shape[1])
    
    # Batch assignment for variables and constraints
    var_batch_el = torch.tensor([0, 0, 0, 1, 1], dtype=torch.long)
    cstr_batch_el = torch.tensor([0, 0, 1], dtype=torch.long)
    
    # Create data object
    data = Data(
        var_feats=var_feats,
        cstr_feats=cstr_feats,
        edge_index=edge_index,
        edge_attr=edge_attr,
        var_batch_el=var_batch_el,
        cstr_batch_el=cstr_batch_el
    )
    
    return data

# Set up arguments for MilpGNN
ARGS = {
    'hidden_dim': (4, 3),
    'n_gnn_layers': 2,
}

# Also provide a sample input to test with
sample_data = create_minimal_data()
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MilpGNN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MilpGNN 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
RegressionHead Ran
ran on a synthesized input · constructed an object; no result was computed
romeov/ml4co-competition/bo_gnn/config/models/baseline.py
code served (permissive licence) · get_code("16b493c43920e35d")
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 an instance of its own class, RegressionHead: the run constructed the object and did not call it, so no result was computed. 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)

class RegressionHead(torch.nn.Module):

    def __init__(self, in_dim=2 * 8, hidden_dim=None, out_dim=2):
        super(RegressionHead, self).__init__()
        if not hidden_dim:
            hidden_dim = 4 * in_dim
        self.out_dim = out_dim
        self.lin1 = torch.nn.Linear(in_features=in_dim, out_features=hidden_dim)
        self.lin2 = torch.nn.Linear(in_features=hidden_dim, out_features=hidden_dim)
        self.lin3_mu = torch.nn.Linear(in_features=hidden_dim, out_features=out_dim)
        self.lin3_sig = torch.nn.Linear(in_features=hidden_dim, out_features=out_dim)

    def forward(self, x):
        x = self.lin1(x).relu_()
        x = self.lin2(x).relu_()
        mu = self.lin3_mu(x)
        sig = self.lin3_sig(x).exp()
        return (mu, sig)


# Create test input tensor: batch_size=2, in_dim=16 (2*8)
test_input = torch.randn(2, 16)

# Create RegressionHead instance with default parameters
regression_head = RegressionHead()

# Execute the forward pass
output = regression_head(test_input)

ARGS = {
    "in_dim": 16,
    "hidden_dim": None,
    "out_dim": 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type RegressionHead — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a RegressionHead 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
ConfigPerformanceRegressor Not yet run
unverified (harvested, no recorded run)
romeov/ml4co-competition/bo_gnn/config/models/baseline.py
code served (permissive licence) · get_code("ce311acee9feef57")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 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 5 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We present our submission for the configuration task of the Machine Learning for Combinatorial Optimization (ML4CO) NeurIPS 2021 competition. The configuration task is to predict a good configuration of the open-source solver SCIP to solve a mixed integer linear program (MILP) efficiently. We pose this task as a supervised learning problem: First, we compile a large dataset of the solver performance for various configurations and all provided MILP instances. Second, we use this data to train a graph neural network that learns to predict a good configuration for a specific instance. The submission was tested on the three problem benchmarks of the competition and improved solver performance over the default by 12% and 35% and 8% across the hidden test instances. We ranked 3rd out of 15 on the global leaderboard and won the student leaderboard. We make our code publicly available at https://github.com/RomeoV/ml4co-competition .

For agents

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

get_harvested_code_for_paper("2202.04910")
get_code_for_paper("2202.04910")
have("2202.04910")

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

curl https://syntology.ai/api/ran/2202.04910.json

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

[![Syntology run record](https://syntology.ai/api/ran/2202.04910.svg)](https://syntology.ai/paper/2202.04910)

Connect an agent — have() is free.