SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2211.00631 · NeurIPS · 2022

Composite Feature Selection Using Deep Ensembles

Pietro Liò, Mihaela Van Der Schaar, Alexander Norcliffe, Fergus Imrie

arXiv · PDF · Open in the Atlas

Code that ran

13 of 13 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, 1 violated it and 11 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/2211.00631.json · badge: /api/ran/2211.00631.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
vanderschaarlab/Composite-Feature-Selection canonical (classifier label)
classifier basis not recorded on the link
7 ran · 0 no recorded run
a-norcliffe/composite-feature-selection canonical (classifier label)
classifier basis: classifier_2025
6 ran · 0 no recorded run
FunctionStatusWhere it lives
CompFS Ran
ran on a synthesized input · constructed an object; no result was computed
a-norcliffe/composite-feature-selection/compfs/models/compfs.py
code served (permissive licence) · get_code("bb117b596f431931")
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, CompFS: 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 as nn
import torch.nn.functional as F
import numpy as np

torch.manual_seed(0)

# Define a simple threshold function
def threshold_func(p):
    return (p > 0.5).float()

# Define a simple loss function
def loss_func(pred, target):
    return F.mse_loss(pred, target)

# Create the config dictionary
ARGS = {
    'config_dict': {
        'nlearners': 3,
        'in_dim': 4,
        'h_dim': 5,
        'out_dim': 2,
        'threshold_func': threshold_func,
        'temp': 0.5,
        'beta_s': 0.1,
        'beta_s_decay': 0.95,
        'beta_d': 0.1,
        'beta_d_decay': 0.95,
        'loss_func': loss_func,
    }
}
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 CompFS — 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 CompFS 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
FullyConnected Ran
ran on a synthesized input · constructed an object; no result was computed
a-norcliffe/composite-feature-selection/compfs/models/compfs.py
code served (permissive licence) · get_code("8ae878fd4771f41b")
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, FullyConnected: 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 as nn

torch.manual_seed(0)

ARGS = {
    "in_dim": 4,
    "h_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 FullyConnected — 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 FullyConnected 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
Gate Ran
ran on a synthesized input · constructed an object; no result was computed
a-norcliffe/composite-feature-selection/compfs/models/compfs.py
code served (permissive licence) · get_code("55b31ad1b5c1ab68")
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, Gate: 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 as nn

torch.manual_seed(0)

# Define a simple threshold function that converts probabilities to binary masks
def threshold_func(p):
    """Convert probabilities to binary masks using a threshold of 0.5"""
    return (p > 0.5).float()

# Define the arguments
ARGS = {
    'in_dim': 4,
    'threshold_func': threshold_func,
    'temp': 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 Gate — 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 Gate 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
SingleFeatureSelector Ran
ran on a synthesized input · constructed an object; no result was computed
a-norcliffe/composite-feature-selection/compfs/models/compfs.py
code served (permissive licence) · get_code("0df3c850588a3e59")
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, SingleFeatureSelector: 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 as nn

torch.manual_seed(0)

# Define a simple threshold function for the gate
def threshold_func(p):
    """Convert probabilities to binary mask using threshold at 0.5"""
    return (p > 0.5).float()

# Define ARGS with all required parameters
ARGS = {
    "in_dim": 4,
    "h_dim": 3,
    "out_dim": 2,
    "threshold_func": threshold_func,
    "temp": 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 SingleFeatureSelector — 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 SingleFeatureSelector 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
accuracy Ran
ran on a synthesized input
vanderschaarlab/Composite-Feature-Selection/compfs/metrics.py
code served (permissive licence) · get_code("96774edcbdcbf02f")
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 torch

torch.manual_seed(0)

# The function takes two parameters:
# x: a tensor that will be passed to torch.argmax(x, dim=-1), so needs shape (batch, ..., num_classes)
# y: a tensor that will be compared with argmax result and passed to len(), so needs to be 1D with batch size

# Create x: logits/predictions with shape (batch=2, num_classes=4)
x = torch.randn(2, 4)

# Create y: ground truth labels with shape (batch=2,)
y = torch.tensor([0, 2])

ARGS = {
    "x": x,
    "y": y
}
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 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
featurise_mols Ran
ran on a synthesized input
vanderschaarlab/Composite-Feature-Selection/compfs/datasets/chem_featuriser.py
code served (permissive licence) · get_code("d4dd30aec90c6ea3")
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 ndarray, ndim 2, dtype family float, shape (2, 4) 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

class MockMol:
    """Mock RDKit mol object with GetSubstructMatch method"""
    def __init__(self, match_indices_list):
        self.match_indices_list = match_indices_list
        self.call_count = 0
    
    def GetSubstructMatch(self, substruct):
        result = self.match_indices_list[self.call_count]
        self.call_count += 1
        return result

class MockSubstruct:
    """Mock RDKit substruct object"""
    pass

# Create mock molecules with different substructure match patterns
# mol 0: matches substruct 0 and 2
mol_0 = MockMol([
    (0, 1),      # matches substruct 0
    (),          # does not match substruct 1
    (2,),        # matches substruct 2
    ()           # does not match substruct 3
])

# mol 1: matches substruct 1 only
mol_1 = MockMol([
    (),          # does not match substruct 0
    (0, 1, 2),   # matches substruct 1
    (),          # does not match substruct 2
    (3,)         # matches substruct 3
])

mols = [mol_0, mol_1]

# Create mock substructs
substruct_list = [MockSubstruct() for _ in range(4)]

ARGS = {
    "mols": mols,
    "substruct_list": substruct_list
}
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 ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
is_array_in_list Ran
ran, violated the contract we drafted
a-norcliffe/composite-feature-selection/compfs/models/compfs.py
code served (permissive licence) · get_code("dfd24b81d29918ff")
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 arrays with distinct sizes
arr1 = np.array([1, 2, 3, 4])
arr2 = np.array([5, 6, 7, 8])
arr3 = np.array([1, 2, 3, 4])  # Same as arr1
arr4 = np.array([9, 10, 11, 12])

# Create a list of arrays
arr_list = [arr1, arr2, arr4]

# Test array that should be found in the list (equals arr1)
test_arr = np.array([1, 2, 3, 4])

ARGS = {
    "arr": test_arr,
    "arr_list": arr_list
}
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
lasso_accuracy Ran
ran on a synthesized input
vanderschaarlab/Composite-Feature-Selection/compfs/metrics.py
code served (permissive licence) · get_code("fbf0f537e8971813")
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 torch

torch.manual_seed(0)

# The function compares x == y element-wise and computes accuracy
# x and y should be tensors of the same shape
# The function calls len(y), so y must support len()
# The function uses torch.sum() on a boolean tensor result

x = torch.tensor([1, 0, 1, 1], dtype=torch.long)
y = torch.tensor([1, 0, 1, 0], dtype=torch.long)

ARGS = {
    "x": x,
    "y": y,
}
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 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
make_lambda_threshold Ran
ran on a synthesized input
vanderschaarlab/Composite-Feature-Selection/compfs/thresholding_functions.py
code served (permissive licence) · get_code("9237a86cdc2f1959")
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 function. 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)

# The function make_lambda_threshold takes a parameter 'lmd' (lambda)
# It converts it to float and returns a function that compares a tensor p
# against a threshold value using torch.full_like

# Valid argument: a numeric value that can be converted to float
lmd_value = 0.5

ARGS = {
    "lmd": lmd_value
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
make_std_threshold Ran
ran on a synthesized input
vanderschaarlab/Composite-Feature-Selection/compfs/thresholding_functions.py
code served (permissive licence) · get_code("c9a3fee6017301d1")
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 function. 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)

ARGS = {
    "nsigma": 2.0
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
make_top_k_threshold Ran
ran on a synthesized input
vanderschaarlab/Composite-Feature-Selection/compfs/thresholding_functions.py
code served (permissive licence) · get_code("94abfde144bc8d32")
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 function. 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)

ARGS = {
    "k": 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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
set_device Ran
ran; our contract draft was wrong, not the code
a-norcliffe/composite-feature-selection/experiments/run_experiment.py
code served (permissive licence) · get_code("4321a42df5ef22c5")
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) and returned device. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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)

# The function takes a single parameter 'x' which is a string
# It checks if x == 'cpu', otherwise treats it as a cuda device index
# The function uses string operations and torch.cuda.is_available()

ARGS = {
    'x': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type device — a device output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a device 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
sklearn_accuracy Ran
ran on a synthesized input
vanderschaarlab/Composite-Feature-Selection/compfs/metrics.py
code served (permissive licence) · get_code("8ee3d6095cd6cdc9")
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 torch
import numpy as np

torch.manual_seed(0)

# x should be a tensor or array of predicted probabilities/scores
# The function applies thresholding at 0.5, so it needs float values
x = torch.tensor([0.2, 0.7, 0.3, 0.9], dtype=torch.float32)

# y should be a tensor or array of true binary labels
y = torch.tensor([0, 1, 0, 1], dtype=torch.long)

ARGS = {
    "x": x,
    "y": y,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — 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

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 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 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.

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

In many real world problems, features do not act alone but in combination with each other. For example, in genomics, diseases might not be caused by any single mutation but require the presence of multiple mutations. Prior work on feature selection either seeks to identify individual features or can only determine relevant groups from a predefined set. We investigate the problem of discovering groups of predictive features without predefined grouping. To do so, we define predictive groups in terms of linear and non-linear interactions between features. We introduce a novel deep learning architecture that uses an ensemble of feature selection models to find predictive groups, without requiring candidate groups to be provided. The selected groups are sparse and exhibit minimum overlap. Furthermore, we propose a new metric to measure similarity between discovered groups and the ground truth. We demonstrate the utility of our model on multiple synthetic tasks and semi-synthetic chemistry datasets, where the ground truth structure is known, as well as an image dataset and a real-world cancer dataset.

For agents

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

get_harvested_code_for_paper("2211.00631")
get_code_for_paper("2211.00631")
have("2211.00631")

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

curl https://syntology.ai/api/ran/2211.00631.json

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

[![Syntology run record](https://syntology.ai/api/ran/2211.00631.svg)](https://syntology.ai/paper/2211.00631)

Connect an agent — have() is free.