SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2306.04201 · ICML · 2023

Improving Hyperparameter Learning under Approximate Inference in Gaussian Process Models

Arno Solin, John St, Rui Li

arXiv · PDF · Open in the Atlas

Code that ran

18 of 19 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; 1 honoured the contract we drafted, 0 violated it and 14 have no contract verdict; 3 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/2306.04201.json · badge: /api/ran/2306.04201.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
aaltoml/improved-hyperparameter-learning canonical (classifier label)
classifier basis: census
13 ran · 1 no recorded run
secondmind-labs/bayesian_benchmarks canonical (classifier label)
classifier basis: census
5 ran · 0 no recorded run
FunctionStatusWhere it lives
adapt_array Ran
ran, honoured the contract we drafted
aaltoml/improved-hyperparameter-learning/python/bayesian_benchmarks/database_utils.py
code served (permissive licence) · get_code("f870491bfd736f8d")
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 memoryview. 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 sqlite3
import numpy as np
import io

# Create a small numpy array to pass to adapt_array
test_array = np.array([[1.0, 2.0, 3.0, 4.0],
                       [5.0, 6.0, 7.0, 8.0],
                       [9.0, 10.0, 11.0, 12.0]])

ARGS = {
    "arr": test_array
}
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 memoryview — a memoryview output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a memoryview 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
add_classficiation Ran
ran; our contract draft was wrong, not the code
aaltoml/improved-hyperparameter-learning/python/bayesian_benchmarks/data.py
code served (permissive licence) · get_code("9b436ef2d0792040")
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 ClassificationDataset. 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 ClassificationDataset:
    """Minimal stand-in class for a classification dataset."""
    def __init__(self, name):
        self.name = name

# Create a simple classification dataset object
dataset = ClassificationDataset(name="test_dataset")

# Define ARGS mapping the parameter to the value
ARGS = {
    "C": dataset
}
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 ClassificationDataset — a ClassificationDataset output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ClassificationDataset 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
add_regression Ran
ran; our contract draft was wrong, not the code
aaltoml/improved-hyperparameter-learning/python/bayesian_benchmarks/data.py
code served (permissive licence) · get_code("5cadf1fc3261f6c0")
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 RegressionDataset. 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 RegressionDataset:
    def __init__(self):
        self.name = "test_regression_dataset"

C = RegressionDataset()

ARGS = {
    "C": C
}
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 RegressionDataset — a RegressionDataset output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a RegressionDataset 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
broadcasting_ABT Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/src/utils/util.py
code served (permissive licence) · get_code("0989691c9ccd9673")
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 EagerTensor, ndim 3, dtype family float, shape (4, 2, 2) 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 tensorflow as tf
import numpy as np

# Create small test tensors
# A and B should both be [M, N] shaped tensors
M = 2  # number of samples
N = 4  # dimension size

# Create A as [M, N] = [2, 4]
A = tf.constant([
    [1.0, 2.0, 3.0, 4.0],
    [5.0, 6.0, 7.0, 8.0]
], dtype=tf.float32)

# Create B as [M, N] = [2, 4]
B = tf.constant([
    [0.1, 0.2, 0.3, 0.4],
    [0.5, 0.6, 0.7, 0.8]
], dtype=tf.float32)

ARGS = {
    "A": A,
    "B": B
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 3, dtype family float, shape (4, 2, 2) — shape (4, 2, 2) 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
broadcasting_ATB Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/src/utils/util.py
code served (permissive licence) · get_code("a2f6bd90423824b5")
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 EagerTensor, ndim 3, dtype family float, shape (4, 1, 1) 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 tensorflow as tf
import numpy as np

# Set seed for reproducibility
tf.random.set_seed(0)
np.random.seed(0)

# Create small test tensors
# A and B should be [M, N] shaped tensors
# M = 2 (batch dimension), N = 4 (feature dimension)
A = tf.constant([[1.0, 2.0, 3.0, 4.0],
                  [5.0, 6.0, 7.0, 8.0]], dtype=tf.float32)  # [2, 4]

B = tf.constant([[0.5, 1.5, 2.5, 3.5],
                  [4.5, 5.5, 6.5, 7.5]], dtype=tf.float32)  # [2, 4]

ARGS = {
    "A": A,
    "B": B
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 3, dtype family float, shape (4, 1, 1) — shape (4, 1, 1) 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
classification_model Ran
ran on a synthesized input
secondmind-labs/bayesian_benchmarks/bayesian_benchmarks/models/non_bayesian_models.py
code served (permissive licence) · get_code("63e11d72c9447f5e")
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 type. 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
from sklearn.ensemble import RandomForestClassifier

# Create a minimal sklearn model that has fit() and predict_proba() methods
sklearn_model = RandomForestClassifier(n_estimators=2, random_state=0)

# Define ARGS with the required parameter
ARGS = {
    'model': sklearn_model
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
conditional_from_precision_sites_white Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/src/utils/util.py
code served (permissive licence) · get_code("710c670579f1dfc9")
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 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 tensorflow as tf
import numpy as np

tf.keras.backend.set_floatx('float64')

M = 4
N = 3
L_dim = 2

Kuu = tf.constant(np.random.randn(M, M), dtype=tf.float64)
Kuu = Kuu @ tf.linalg.matrix_transpose(Kuu) + tf.eye(M, dtype=tf.float64)

Kff = tf.constant(np.random.randn(N, 1), dtype=tf.float64)

Kuf = tf.constant(np.random.randn(M, N), dtype=tf.float64)

l = tf.constant(np.random.randn(M, L_dim), dtype=tf.float64)

L_mat = tf.constant(np.random.randn(L_dim, M, M), dtype=tf.float64)

ARGS = {
    "Kuu": Kuu,
    "Kff": Kff,
    "Kuf": Kuf,
    "l": l,
    "L": L_mat,
    "L2": None,
    "jitter": 1e-09
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 3, 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
convert_array Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/python/bayesian_benchmarks/database_utils.py
code served (permissive licence) · get_code("6969043e1e929528")
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 1, dtype family float, 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 numpy as np
import io

# Create a valid numpy array and serialize it to bytes
array_data = np.array([1, 2, 3, 4, 5], dtype=np.float32)
buffer = io.BytesIO()
np.save(buffer, array_data)
text = buffer.getvalue()

ARGS = {
    "text": text
}
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 1, dtype family float, 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
dict_to_lists Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/python/bayesian_benchmarks/database_utils.py
code served (permissive licence) · get_code("287775627678ecfc")
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)

# The function takes a dict and flattens it into keys and values
# It uses: d.keys() and d[k] for each key
# So we need a simple dict with some key-value pairs

ARGS = {
    "d": {
        "key1": 42,
        "key2": "value2",
        "key3": [1, 2, 3],
        "key4": torch.tensor([1.0, 2.0, 3.0, 4.0])
    }
}
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 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
dict_to_lists Ran
ran on a synthesized input
secondmind-labs/bayesian_benchmarks/bayesian_benchmarks/database_utils.py
code served (permissive licence) · get_code("fc15d16c3abea6c5")
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)

# The function dict_to_lists takes a dict as input
# It accesses d.keys() and d[k] for each key
# So we need a simple dict with various value types

ARGS = {
    "d": {
        "key1": 42,
        "key2": "value2",
        "key3": [1, 2, 3],
        "key4": torch.tensor([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]]),
        "key5": {"nested": "dict"}
    }
}
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
expand_to_absolute Ran
ran on a synthesized input
secondmind-labs/bayesian_benchmarks/bayesian_benchmarks/paths.py
code served (permissive licence) · get_code("c4378427f9c3bf42")
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 os

def expand_to_absolute(path):
    if './' == path[:2]:
        return os.path.join(os.path.dirname(os.path.realpath(__file__)), path[2:])
    else:
        return path

ARGS = {
    'path': './test_file.txt'
}
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 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
gpflow_log_inv_probit Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/src/new_bernoulli/compare_logphi.py
code served (permissive licence) · get_code("8f109eb7cd08f71b")
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 EagerTensor, ndim 1, dtype family float, shape (201,) 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
import tensorflow as tf

# Create a small numpy array as input
x_input = np.linspace(-10, 10, 201).astype(np.float64)

# Define jitter parameter
jitter_input = np.float64(0.001)

# Create the ARGS dictionary with all parameters for gpflow_log_inv_probit
ARGS = {
    "x": x_input,
    "jitter": jitter_input
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 1, dtype family float, shape (201,) — shape (201,) 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
inv_probit Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/src/new_bernoulli/compare_logphi.py
code served (permissive licence) · get_code("b96b3647dd98a6d3")
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 EagerTensor, ndim 1, dtype family float, shape (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
import tensorflow as tf

x = np.linspace(-10, 10, 201)

# Create arguments for inv_probit function
x_input = np.array([-2.5, 0.0, 2.5, 5.0], dtype=np.float64)
jitter_value = np.float64(0.001)

ARGS = {
    "x": x_input,
    "jitter": jitter_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 1, dtype family float, shape (4,) — shape (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
logphi Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/src/new_bernoulli/logphi.py
code served (permissive licence) · get_code("6de929b77412ac69")
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 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 numpy as np
import tensorflow as tf

# Create test input for logphi function
# The function expects z to be a tensor-like object that supports operations
z_test = tf.constant([-0.5, 0.1, -12.0, 5.0], dtype=tf.float64)

ARGS = {
    "z": z_test,
    "grad": True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, 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
new_log_inv_probit Ran
ran on a synthesized input
aaltoml/improved-hyperparameter-learning/src/new_bernoulli/logphi.py
code served (permissive licence) · get_code("f328d35a5192c364")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

# Create small test input
z = tf.constant([[-2.5, -1.0, 0.5, 2.0], 
                 [1.5, -0.5, 0.0, 3.0]], dtype=tf.float64)

ARGS = {
    'z': z
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
non_bayesian_model Ran
ran on a synthesized input
secondmind-labs/bayesian_benchmarks/bayesian_benchmarks/models/non_bayesian_models.py
code served (permissive licence) · get_code("f21ba8b805fe6210")
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 type. 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
from sklearn import linear_model

# The function takes two parameters: name and task
# name: a string identifying the model type
# task: a string identifying the task type ('regression' or 'classification')

ARGS = {
    'name': 'linear',
    'task': 'regression'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
normalize Ran
ran; our contract draft was wrong, not the code
aaltoml/improved-hyperparameter-learning/python/bayesian_benchmarks/data.py
code served (permissive licence) · get_code("cba0faabe1fa7dc7")
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 numpy as np

# Create a small numpy array with shape (2, 4) - batch size 2, feature dimension 4
X = np.random.randn(2, 4).astype(np.float32)

ARGS = {
    "X": X
}
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 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
regression_model Ran
ran on a synthesized input
secondmind-labs/bayesian_benchmarks/bayesian_benchmarks/models/non_bayesian_models.py
code served (permissive licence) · get_code("8bd09fd901fa3107")
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 type. 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
from sklearn.linear_model import LinearRegression

# Create a simple scikit-learn model that has fit() and predict() methods
model = LinearRegression()

ARGS = {
    "model": model
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
np_logphi Not yet run
unverified (harvested, no recorded run)
aaltoml/improved-hyperparameter-learning/src/new_bernoulli/logphi.py
code served (permissive licence) · get_code("485726aef7679d8c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 11 are also present in other papers whose code we hold (11 with at least one paper dated earlier, 0 the same month, 3 later) and 7 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
adapt_array 2 other papers
2 dated earlier
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
2301.11325 (2023-01)
MusicLM: Generating Music From Text
add_classficiation 2 other papers
1 dated earlier, 1 later
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
2410.05757 (2024-10)
Temperature Optimization for Bayesian Deep Learning
add_regression 2 other papers
1 dated earlier, 1 later
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
2410.05757 (2024-10)
Temperature Optimization for Bayesian Deep Learning
normalize 2 other papers
1 dated earlier, 1 later
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
2410.05757 (2024-10)
Temperature Optimization for Bayesian Deep Learning
classification_model 1 other paper
1 dated earlier
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
conditional_from_precision_sites_white 1 other paper
1 dated earlier
2111.03412 (2021-11)
Dual Parameterization of Sparse Variational Gaussian Processes
convert_array 1 other paper
1 dated earlier
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
dict_to_lists 1 other paper
1 dated earlier
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
expand_to_absolute 1 other paper
1 dated earlier
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
non_bayesian_model 1 other paper
1 dated earlier
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
regression_model 1 other paper
1 dated earlier
1806.05975 (2018-06)
Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors

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

Approximate inference in Gaussian process (GP) models with non-conjugate likelihoods gets entangled with the learning of the model hyperparameters. We improve hyperparameter learning in GP models and focus on the interplay between variational inference (VI) and the learning target. While VI's lower bound to the marginal likelihood is a suitable objective for inferring the approximate posterior, we show that a direct approximation of the marginal likelihood as in Expectation Propagation (EP) is a better learning objective for hyperparameter optimization. We design a hybrid training procedure to bring the best of both worlds: it leverages conjugatecomputation VI for inference and uses an EP-like marginal likelihood approximation for hyperparameter learning. We compare VI, EP, Laplace approximation, and our proposed training procedure and empirically demonstrate the effectiveness of our proposal across a wide range of data sets.

For agents

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

get_harvested_code_for_paper("2306.04201")
get_code_for_paper("2306.04201")
have("2306.04201")

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

curl https://syntology.ai/api/ran/2306.04201.json

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

[![Syntology run record](https://syntology.ai/api/ran/2306.04201.svg)](https://syntology.ai/paper/2306.04201)

Connect an agent — have() is free.