SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1610.02136 · 2016

A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks

Dan Hendrycks, Kevin Gimpel

arXiv · PDF · Open in the Atlas

Code that ran

19 of 21 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; 4 honoured the contract we drafted, 0 violated it and 11 have no contract verdict; 4 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/1610.02136.json · badge: /api/ran/1610.02136.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
hendrycks/error-detection canonical (classifier label)
classifier basis: census
12 ran · 2 no recorded run
2sang/OOD-baseline community
Papers with Code archive
2 ran · 0 no recorded run
kingjamessong/rankfeat community
Papers with Code archive
1 ran · 0 no recorded run
lancopku/avg-avg community
Papers with Code archive
1 ran · 0 no recorded run
oliverzhang42/ood_medical_images community
Papers with Code archive
1 ran · 0 no recorded run
guyAmit/GLOD community
Papers with Code archive
1 ran · 0 no recorded run
kobybibas/pnml_ood_detection community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
clipped_gelu Ran
ran on a synthesized input
hendrycks/error-detection/ASR/CTC/CTC_eval.py
code served (permissive licence) · get_code("e70a9b1d04bacb16")
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 SymbolicTensor. 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 tensorflow as tf
import numpy as np

def clipped_gelu(x):
    return tf.minimum(0.5 * x * (1 + tf.tanh(x)), 6)

# Create a small test tensor
x = tf.constant(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, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type SymbolicTensor — a SymbolicTensor 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
data_lists_to_batches Ran
ran on a synthesized input
hendrycks/error-detection/ASR/CTC/utils.py
code served (permissive licence) · get_code("5f1003adf845195e")
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

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

# Create small input list: list of 2-d numpy arrays with dimensions nFeatures x timesteps
# Using distinct sizes: nFeatures=4, variable timesteps
inputList = [
    np.random.randn(4, 5),  # 4 features, 5 timesteps
    np.random.randn(4, 7),  # 4 features, 7 timesteps
    np.random.randn(4, 6),  # 4 features, 6 timesteps
    np.random.randn(4, 8),  # 4 features, 8 timesteps
]

# Create small target list: list of 1-d arrays/lists of ints
# These represent character indices or token IDs
targetList = [
    np.array([1, 2, 3]),
    np.array([2, 3]),
    np.array([1, 2, 3, 4]),
    np.array([3, 4, 5]),
]

# Batch size
batchSize = 2

ARGS = {
    "inputList": inputList,
    "targetList": targetList,
    "batchSize": batchSize,
}
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
data_to_mat Ran
ran on a synthesized input
hendrycks/error-detection/Appendix_B_Cautious_Classification/POS/helper_functions_twitter.py
code served (permissive licence) · get_code("70ffbe91cbb5596b")
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
import io
import tempfile
import os

# Create a temporary file with sample data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt', encoding='utf-8')
temp_filename = temp_file.name

# Write sample training data: word\tlabel format, blank lines separate tweets
sample_data = """apple\tNOUN
is\tVERB
red\tADJ

the\tDET
cat\tNOUN
sat\tVERB

"""

temp_file.write(sample_data)
temp_file.close()

# Define vocab - list of strings
vocab = ['apple', 'is', 'red', 'the', 'cat', 'sat', 'UUUNKKK', '</s>']

# Define tag_to_number - dictionary mapping tags to numerical indices
tag_to_number = {
    'NOUN': 0,
    'VERB': 1,
    'ADJ': 2,
    'DET': 3
}

ARGS = {
    'filename': temp_filename,
    'vocab': vocab,
    'tag_to_number': tag_to_number,
    'window_size': 1,
    'start_symbol': 'UUUNKKK',
    'one_hot': False,
    'return_labels': True
}
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
embeddings_to_dict Ran
ran on a synthesized input
hendrycks/error-detection/Appendix_B_Cautious_Classification/POS/helper_functions_twitter.py
code served (permissive licence) · get_code("7d646d3612cadf34")
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 numpy as np
import io
import tempfile
import os

# Create a temporary file with word embeddings in the expected format
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt', encoding='utf-8')
temp_filename = temp_file.name

# Write sample embeddings data in the format: word[tab]dim1[space]dim2[space]...[space]dim50
embeddings_data = [
    "hello\t0.1 0.2 0.3 0.4 0.5",
    "world\t0.6 0.7 0.8 0.9 1.0",
    "python\t1.1 1.2 1.3 1.4 1.5"
]

for line in embeddings_data:
    temp_file.write(line + '\n')

temp_file.close()

ARGS = {
    'filename': temp_filename
}
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 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
entropy_from_distribution Ran
ran, honoured the contract we drafted
2sang/OOD-baseline/exp_util.py
pointer only (licence: NONE) · get_code("2ae461db08765c9b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, 2sang/OOD-baseline when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

entropy_stats Ran
ran; our contract draft was wrong, not the code
2sang/OOD-baseline/exp_util.py
pointer only (licence: NONE) · get_code("72903b57f29a3ee1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, 2sang/OOD-baseline when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_Mahalanobis_score Ran
ran; our contract draft was wrong, not the code
kingjamessong/rankfeat/utils/mahalanobis_lib.py
pointer only (licence: NONE) · get_code("d2b7861b22672799")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kingjamessong/rankfeat when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_base_score Ran
ran, honoured the contract we drafted
lancopku/avg-avg/lib/inference/base.py
code served (permissive licence) · get_code("d47e87d197d79e74")
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 ndarray, 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 torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

# Create a minimal model that satisfies the function's usage
class MinimalModel(nn.Module):
    def __init__(self, num_classes=3):
        super().__init__()
        self.linear = nn.Linear(4, num_classes)
    
    def forward(self, input_ids, attention_mask=None):
        # Return an object with a logits attribute
        logits = self.linear(input_ids)
        
        class Output:
            pass
        
        output = Output()
        output.logits = logits
        return output

# Create a minimal model instance
model = MinimalModel(num_classes=3)

# Create a minimal data loader
# Each batch should be (input_ids, labels, attention_masks)
data_loader = [
    (
        torch.randn(2, 4),  # input_ids: batch_size=2, seq_len=4
        torch.tensor([0, 1]),  # labels: batch_size=2
        torch.ones(2, 4)  # attention_masks: batch_size=2, seq_len=4
    ),
    (
        torch.randn(2, 4),
        torch.tensor([2, 0]),
        torch.ones(2, 4)
    )
]

ARGS = {
    "model": model,
    "data_loader": data_loader
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch, tqdm — 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 (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
get_confidence_baseline Ran
ran; our contract draft was wrong, not the code
oliverzhang42/ood_medical_images/src/utils.py
pointer only (licence: NONE) · get_code("0d73cb6aa8f3382b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, oliverzhang42/ood_medical_images when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

load_batch Ran
ran on a synthesized input
hendrycks/error-detection/Vision/load_cifar10.py
code served (permissive licence) · get_code("8633f606824fb007")
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 pickle
import tempfile
import os

# Create a temporary file with pickled data
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, 'batch.pkl')

# Create sample data that matches what load_batch expects
sample_data = {
    'data': [[1, 2, 3, 4], [5, 6, 7, 8]],
    'labels': [0, 1]
}

# Write the pickle file
with open(temp_file, 'wb') as f:
    pickle.dump(sample_data, f)

ARGS = {
    'fpath': temp_file
}
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 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
load_batch Ran
ran on a synthesized input
hendrycks/error-detection/Vision/load_cifar100.py
code served (permissive licence) · get_code("4e5831192b762838")
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 pickle
import tempfile
import os

# Create a temporary pickle file with the expected structure
temp_dir = tempfile.mkdtemp()
fpath = os.path.join(temp_dir, "batch.pkl")

# Create a minimal valid pickle file with the expected dictionary structure
data_dict = {
    'data': [[1, 2, 3, 4], [5, 6, 7, 8]],
    'fine_labels': [0, 1],
    'coarse_labels': [0, 0]
}

with open(fpath, 'wb') as f:
    pickle.dump(data_dict, f)

ARGS = {
    'fpath': fpath,
    'label_mode': 'fine'
}
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 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
load_batched_data Ran
ran on a synthesized input
hendrycks/error-detection/ASR/CTC/utils.py
code served (permissive licence) · get_code("201106667cf3cb5c")
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
import tempfile
import os

# Create temporary directories for specPath and targetPath
spec_dir = tempfile.mkdtemp()
target_dir = tempfile.mkdtemp()

# Generate some sample spec files (2D numpy arrays: nFeatures x timesteps)
nFeatures = 4
for i in range(3):
    spec = np.random.randn(nFeatures, 5 + i)  # Different timesteps for each
    np.save(os.path.join(spec_dir, f'spec_{i}.npy'), spec)

# Generate some sample target files (1D arrays of integers)
for i in range(3):
    target = np.array([1, 2, 3, 1], dtype=np.int32)  # Example target sequence
    np.save(os.path.join(target_dir, f'target_{i}.npy'), target)

batchSize = 2

ARGS = {
    'specPath': spec_dir,
    'targetPath': target_dir,
    'batchSize': batchSize
}
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
predict_llr Ran
ran, honoured the contract we drafted
guyAmit/GLOD/FOOD/Gaussain_layer.py
pointer only (licence: NONE) · get_code("0222c378f0e6e52e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, guyAmit/GLOD when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

softmax Ran
ran, honoured the contract we drafted
kobybibas/pnml_ood_detection/src/score_utils.py
pointer only (licence: NONE) · get_code("7ddc824e88e78020")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, kobybibas/pnml_ood_detection when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

softmax Ran
ran on a synthesized input
hendrycks/error-detection/ASR/CTC/CTC_eval.py
code served (permissive licence) · get_code("f3dea44d35e56a40")
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

# The function takes one parameter 'x' which is used as:
# - np.exp(x - ...) : needs to be array-like
# - np.max(x, axis=1, keepdims=True) : needs to support axis=1, so 2D array
# - np.sum(e_x, axis=1, keepdims=True) : confirms 2D structure (batch, features)

# Create a small 2D numpy array: batch_size=2, features=4
x = np.array([
    [1.0, 2.0, 3.0, 4.0],
    [0.5, 1.5, 2.5, 3.5]
], dtype=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 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
target_list_to_sparse_tensor Ran
ran on a synthesized input
hendrycks/error-detection/ASR/CTC/utils.py
code served (permissive licence) · get_code("907b6771b97cb718")
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

# Construct a valid targetList: a list of sequences where each sequence
# contains integer values (like character indices for ASR)
targetList = [
    [1, 2, 3],           # First target sequence
    [4, 5],              # Second target sequence
    [0, 1, 2, 3, 4]      # Third target sequence
]

ARGS = {
    "targetList": targetList
}
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
to_categorical Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
hendrycks/error-detection/Vision/load_cifar10.py
code served (permissive licence) · get_code("70f5b5175a23f59d")
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 (5, 3) 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

# The function to_categorical takes:
# - y: array-like of class indices
# - nb_classes: number of classes (int or None/falsy)

# Create a small sample of class indices
y_sample = np.array([0, 1, 2, 1, 0])

# Set number of classes
nb_classes_value = 3

ARGS = {
    'y': y_sample,
    'nb_classes': nb_classes_value
}
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 (5, 3) — shape (5, 3) 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
unpickle Ran
ran on a synthesized input
hendrycks/error-detection/Vision/CIFAR_Detection.py
code served (permissive licence) · get_code("9cf114efaf65dc13")
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 pickle
import tempfile
import os

# Create a temporary pickle file for the unpickle function to read
temp_dir = tempfile.mkdtemp()
test_data = {"key1": "value1", "key2": [1, 2, 3], "key3": {"nested": "dict"}}
test_file = os.path.join(temp_dir, "test.pkl")

with open(test_file, 'wb') as f:
    pickle.dump(test_data, f)

ARGS = {
    "file": test_file
}
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 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
word_list_to_embedding Ran
ran on a synthesized input
hendrycks/error-detection/Appendix_B_Cautious_Classification/POS/helper_functions_twitter.py
code served (permissive licence) · get_code("c08d1e05487f7323")
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, 250) 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

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

# Create a small word matrix: 2 samples, 5 words per sample (window_size=2)
# Values are word indices or word strings
words = np.array([
    ['the', 'quick', 'brown', 'fox', 'jumps'],
    ['over', 'the', 'lazy', 'dog', 'runs']
], dtype=object)

# Create a small embedding dictionary
# Each word maps to an embedding of dimension 50
embeddings = {
    'the': np.random.randn(50).astype(np.float32),
    'quick': np.random.randn(50).astype(np.float32),
    'brown': np.random.randn(50).astype(np.float32),
    'fox': np.random.randn(50).astype(np.float32),
    'jumps': np.random.randn(50).astype(np.float32),
    'over': np.random.randn(50).astype(np.float32),
    'lazy': np.random.randn(50).astype(np.float32),
    'dog': np.random.randn(50).astype(np.float32),
    'runs': np.random.randn(50).astype(np.float32),
}

embedding_dimension = 50

ARGS = {
    'words': words,
    'embeddings': embeddings,
    'embedding_dimension': embedding_dimension,
}
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, 250) — shape (2, 250) 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
load_data10 Not yet run
unverified (harvested, no recorded run)
hendrycks/error-detection/Vision/load_cifar10.py
code served (permissive licence) · get_code("9db68167799bea76")
How it ran

no invocation recorded.

load_data100 Not yet run
unverified (harvested, no recorded run)
hendrycks/error-detection/Vision/load_cifar100.py
code served (permissive licence) · get_code("634643388318c5b7")
How it ran

no invocation recorded.

Also present in other papers' code

Of 21 samples lifted from this paper's linked repositories, 21 carry a function identity and 0 do not. Of the 21 distinct functions with an identity, 6 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 0 the same month, 2 later) and 15 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
to_categorical 3 other papers
2 dated earlier, 1 later
1606.08415 (2016-06)
Gaussian Error Linear Units (GELUs)
1608.00530 (2016-08)
Early Methods for Detecting Adversarial Images
aaai_28047 (2024 (year only))
Rethinking Robustness of Model Attributions
get_base_score
brick of this paper's assembly: entry
2 other papers
2 later
2022.findings-emnlp.497 (2022 (year only))
Holistic Sentence Embeddings for Better Out-of-Distribution Detection
2301.07597 (2023-01)
How Close is ChatGPT to Human Experts? Comparison Corpus, Evaluation, and Detection
load_batch 2 other papers
2 dated earlier
1606.08415 (2016-06)
Gaussian Error Linear Units (GELUs)
1608.00530 (2016-08)
Early Methods for Detecting Adversarial Images
load_data10 2 other papers
2 dated earlier
1606.08415 (2016-06)
Gaussian Error Linear Units (GELUs)
1608.00530 (2016-08)
Early Methods for Detecting Adversarial Images
unpickle 2 other papers
2 dated earlier
1606.08415 (2016-06)
Gaussian Error Linear Units (GELUs)
1608.00530 (2016-08)
Early Methods for Detecting Adversarial Images
embeddings_to_dict 1 other paper
1 dated earlier
1606.08415 (2016-06)
Gaussian Error Linear Units (GELUs)

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 consider the two related problems of detecting if an example is misclassified or out-of-distribution. We present a simple baseline that utilizes probabilities from softmax distributions. Correctly classified examples tend to have greater maximum softmax probabilities than erroneously classified and out-of-distribution examples, allowing for their detection. We assess performance by defining several tasks in computer vision, natural language processing, and automatic speech recognition, showing the effectiveness of this baseline across all. We then show the baseline can sometimes be surpassed, demonstrating the room for future research on these underexplored detection tasks.

For agents

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

get_harvested_code_for_paper("1610.02136")
get_code_for_paper("1610.02136")
have("1610.02136")

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

curl https://syntology.ai/api/ran/1610.02136.json

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

[![Syntology run record](https://syntology.ai/api/ran/1610.02136.svg)](https://syntology.ai/paper/1610.02136)

Connect an agent — have() is free.