SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1908.08962 · 2019

Well-Read Students Learn Better: On the Importance of Pre-training Compact Models

arXiv · PDF · Open in the Atlas

Code that ran

21 of 32 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 10 have no contract verdict; 10 failures of our instrument; 11 no recorded run), measured 2026-11-21; 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/1908.08962.json · badge: /api/ran/1908.08962.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
Arthurizijar/Bert_Airport community (classifier label)
classifier basis not recorded on the link
7 ran · 3 no recorded run
somiltg/bert community (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
coronazap/bert_client community (classifier label)
classifier basis not recorded on the link
2 ran · 2 no recorded run
TonyX19/bert_hyperpartisan community (classifier label)
classifier basis not recorded on the link
2 ran · 0 no recorded run
StoneGH/bert community (classifier label)
classifier basis not recorded on the link
2 ran · 0 no recorded run
the-Quert/iNLPfun community (classifier label)
classifier basis not recorded on the link
1 ran · 1 no recorded run
Maz101/Bert community (classifier label)
classifier basis not recorded on the link
1 ran · 1 no recorded run
paolanu/BERT_epitope community (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
SpikeKing/My-Bert community (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
halo090770/bert community (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
Chonwai/Learning_BERT community (classifier label)
classifier basis not recorded on the link
0 ran · 3 no recorded run
MOHAMEDELDAKDOUKY/bert_adjusted community (classifier label)
classifier basis not recorded on the link
0 ran · 1 no recorded run
FunctionStatusWhere it lives
convert_example_to_feature Ran
ran on a synthesized input
paolanu/BERT_epitope/convert_examples_to_features.py
code served (permissive licence) · get_code("b4a25fd2a133d044")
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 InputFeatures. 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 numpy as np

class SimpleTokenizer:
    def __init__(self):
        self.vocab = {
            '[CLS]': 101,
            '[SEP]': 102,
            'hello': 1,
            'world': 2,
            'test': 3,
            'sample': 4,
            'text': 5,
        }
    
    def tokenize(self, text):
        if text is None:
            return []
        words = text.lower().split()
        return [w if w in self.vocab else 'test' for w in words]
    
    def convert_tokens_to_ids(self, tokens):
        return [self.vocab.get(token, 0) for token in tokens]

class SimpleExample:
    def __init__(self, text_a, text_b=None, label=None):
        self.text_a = text_a
        self.text_b = text_b
        self.label = label

example = SimpleExample(
    text_a="hello world",
    text_b="test sample",
    label="positive"
)

label_map = {
    "positive": 1,
    "negative": 0
}

max_seq_length = 16
tokenizer = SimpleTokenizer()
output_mode = "classification"

example_row = (example, label_map, max_seq_length, tokenizer, output_mode)

ARGS = {
    "example_row": example_row
}
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 InputFeatures — a InputFeatures output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a InputFeatures 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
convert_to_unicode Ran
ran; our contract draft was wrong, not the code
Arthurizijar/Bert_Airport/tokenization.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from huawei-noah/Pretrained-Language-Model, is pointer only (licence: NONE) · get_code("1923fc05163d207d")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from huawei-noah/Pretrained-Language-Model, and GitHub detected no licence in that repository 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.

create_float_feature Ran
ran; our contract draft was wrong, not the code
Arthurizijar/Bert_Airport/create_pretraining_data.py
code served (permissive licence) · get_code("b24b3b9c700da839")
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 Feature. 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 create_float_feature(values):
    feature = tf.train.Feature(float_list=tf.train.FloatList(value=list(values)))
    return feature

# The function takes 'values' parameter and converts it to a list, then creates a FloatList
# We need to provide an iterable of numeric values

values = [1.0, 2.5, 3.7, 4.2]

ARGS = {
    "values": values
}
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 Feature — a Feature output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Feature 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
create_int_feature Ran
ran; our contract draft was wrong, not the code
Arthurizijar/Bert_Airport/create_pretraining_data.py
code served (permissive licence) · get_code("dd3ef48a15aec0cd")
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 Feature. 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

# The function create_int_feature takes 'values' as a parameter
# It converts values to a list and passes it to tf.train.Int64List
# So values should be an iterable of integers

ARGS = {
    "values": [1, 2, 3, 4, 5]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Feature — a Feature output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Feature 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
create_optimizer Ran
ran on a synthesized input
SpikeKing/My-Bert/optimization.py
code served (permissive licence) · get_code("021cbaa59ced6dac")
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 Operation. 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.compat.v1 as tf
import numpy as np

# Disable eager execution for TF1 compatibility
tf.disable_eager_execution()

# Create a simple computational graph with a loss
with tf.Graph().as_default():
    # Create simple variables to form a loss
    x = tf.get_variable(name='x', shape=[2, 4], dtype=tf.float32, initializer=tf.ones_initializer())
    y = tf.get_variable(name='y', shape=[4, 3], dtype=tf.float32, initializer=tf.ones_initializer())
    
    # Compute a simple loss
    z = tf.matmul(x, y)
    loss = tf.reduce_sum(z)
    
    # Define arguments for create_optimizer
    ARGS = {
        'loss': loss,
        'init_lr': 0.001,
        'num_train_steps': 100,
        'num_warmup_steps': 10,
        'use_tpu': False
    }
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 Operation — a Operation output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Operation 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
create_optimizer Ran
ran on a synthesized input
halo090770/bert/optimization_1.py
code served (permissive licence) · get_code("5f1c73e253f08c93")
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 Operation. 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

# Enable TF1 behavior
tf.compat.v1.disable_eager_execution()

# Create a simple computational graph with a loss and variables
tf.compat.v1.reset_default_graph()

# Create some dummy trainable variables
with tf.compat.v1.variable_scope("model"):
    w1 = tf.compat.v1.get_variable(
        "weights_1",
        shape=[4, 3],
        dtype=tf.float32,
        initializer=tf.compat.v1.ones_initializer()
    )
    b1 = tf.compat.v1.get_variable(
        "bias_1",
        shape=[3],
        dtype=tf.float32,
        initializer=tf.compat.v1.zeros_initializer()
    )
    ln = tf.compat.v1.get_variable(
        "LayerNorm_scale",
        shape=[3],
        dtype=tf.float32,
        initializer=tf.compat.v1.ones_initializer()
    )

# Create a simple loss as a function of the variables
x = tf.constant([[1.0, 2.0, 3.0, 4.0], [5.0, 6.0, 7.0, 8.0]], dtype=tf.float32)
y_true = tf.constant([[1.0, 0.0, 1.0], [0.0, 1.0, 1.0]], dtype=tf.float32)
y_pred = tf.matmul(x, w1) + b1
loss = tf.reduce_mean(tf.square(y_pred - y_true))

# Define arguments for create_optimizer
ARGS = {
    "loss": loss,
    "init_lr": 0.001,
    "num_train_steps": 100,
    "num_warmup_steps": 10,
    "use_tpu": False
}
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 Operation — a Operation output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Operation 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
extractLabelXml Ran
ran on a synthesized input
TonyX19/bert_hyperpartisan/seperate.py
code served (permissive licence) · get_code("06627935dcb8a0fe")
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 list, ndim 1, 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 xml.etree.ElementTree as ET
import tempfile
import os

# Create a temporary XML file with the expected structure
xml_content = """<?xml version="1.0" encoding="UTF-8"?>
<root>
    <article id="article1" hyperpartisan="true">
        <title>Test Article 1</title>
    </article>
    <article id="article2" hyperpartisan="false">
        <title>Test Article 2</title>
    </article>
    <article id="article3" hyperpartisan="true">
        <title>Test Article 3</title>
    </article>
    <article id="article4" hyperpartisan="false">
        <title>Test Article 4</title>
    </article>
</root>
"""

# Write to temporary file
temp_dir = tempfile.gettempdir()
xml_file_path = os.path.join(temp_dir, "test_articles.xml")
with open(xml_file_path, 'w') as f:
    f.write(xml_content)

ARGS = {
    'xml_file': xml_file_path
}
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 list, ndim 1, dtype family float — a list 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
gelu Ran
ran, honoured the contract we drafted
Arthurizijar/Bert_Airport/modeling.py
code served (permissive licence) · get_code("d08f324f950148de")
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 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 numpy as np
import tensorflow as tf

# Create a small TensorFlow tensor for testing
x = tf.constant([[-0.5, 0.0, 0.5, 1.0], [0.2, -0.3, 0.8, -0.1]], dtype=tf.float32)

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
get_activation Ran
ran; our contract draft was wrong, not the code
Arthurizijar/Bert_Airport/modeling.py
code served (permissive licence) · get_code("5842ee67106321d6")
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 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 numpy as np
import six
import tensorflow as tf

def gelu(x):
    """Gaussian Error Linear Unit.

  This is a smoother version of the RELU.
  Original paper: https://arxiv.org/abs/1606.08415
  Args:
    x: float Tensor to perform activation.

  Returns:
    `x` with the GELU activation applied.
  """
    cdf = 0.5 * (1.0 + tf.tanh(np.sqrt(2 / np.pi) * (x + 0.044715 * tf.pow(x, 3))))
    return x * cdf

def get_activation(activation_string):
    """Maps a string to a Python function, e.g., "relu" => `tf.nn.relu`.

  Args:
    activation_string: String name of the activation function.

  Returns:
    A Python function corresponding to the activation function. If
    `activation_string` is None, empty, or "linear", this will return None.
    If `activation_string` is not a string, it will return `activation_string`.

  Raises:
    ValueError: The `activation_string` does not correspond to a known
      activation.
  """
    if not isinstance(activation_string, six.string_types):
        return activation_string
    if not activation_string:
        return None
    act = activation_string.lower()
    if act == 'linear':
        return None
    elif act == 'relu':
        return tf.nn.relu
    elif act == 'gelu':
        return gelu
    elif act == 'tanh':
        return tf.tanh
    else:
        raise ValueError('Unsupported activation: %s' % act)


ARGS = {
    'activation_string': 'relu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, six, tensorflow — 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
get_tasks Ran
ran on a synthesized input
the-Quert/iNLPfun/code/download_glue_data.py
code served (permissive licence) · get_code("c54b935b72798660")
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 list, ndim 1, 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 torch

ARGS = {
    'task_names': 'CoLA,SST,MRPC'
}
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 list, ndim 1, dtype family float — a list 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
get_tokens Ran
ran; our contract draft was wrong, not the code
somiltg/bert/evaluate-v2.0.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from saptarshi059/cdqa-project, is pointer only (licence: NONE) · get_code("aa6bc32d9c17a2f1")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from saptarshi059/cdqa-project, and GitHub detected no licence in that repository 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.

init_predict_var Ran
ran on a synthesized input
StoneGH/bert/freeze_graph.py
code served (permissive licence) · get_code("5edf8fc5e1137648")
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 1, 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 os
import pickle
import tempfile

# Create a temporary directory for testing
temp_dir = tempfile.mkdtemp()

# Create a label2id.pkl file in the temporary directory
label2id_data = {'negative': 0, 'positive': 1}
label2id_file = os.path.join(temp_dir, 'label2id.pkl')
with open(label2id_file, 'wb') as wf:
    pickle.dump(label2id_data, wf)

ARGS = {
    'path': temp_dir
}
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, ndim 1, 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
input_fn_builder Ran
ran on a synthesized input
Arthurizijar/Bert_Airport/extract_features.py
code served (permissive licence) · get_code("47a3c792cebb6603")
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 tensorflow as tf

class Feature:
    def __init__(self, unique_id, input_ids, input_mask, input_type_ids):
        self.unique_id = unique_id
        self.input_ids = input_ids
        self.input_mask = input_mask
        self.input_type_ids = input_type_ids

seq_length = 4
num_examples = 2

features = [
    Feature(
        unique_id=0,
        input_ids=[101, 2054, 2003, 102],
        input_mask=[1, 1, 1, 1],
        input_type_ids=[0, 0, 0, 0]
    ),
    Feature(
        unique_id=1,
        input_ids=[101, 1045, 2572, 102],
        input_mask=[1, 1, 1, 1],
        input_type_ids=[0, 0, 0, 0]
    )
]

ARGS = {
    'features': features,
    'seq_length': seq_length
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — 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
loadText Ran
ran on a synthesized input
TonyX19/bert_hyperpartisan/seperate.py
code served (permissive licence) · get_code("ce566df49585f2b5")
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 tempfile
import os

# Create a temporary file with test data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write("id1\tThis is the first sentence.\n")
temp_file.write("id2\tThis is the second sentence.\n")
temp_file.write("id1\tThis is another sentence for id1.\n")
temp_file.write("id3\tThis is the third sentence.\n")
temp_file.close()

ARGS = {
    'text_dir': temp_file.name
}
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
load_vocab Ran
ran on a synthesized input
Maz101/Bert/tokenization.py
code served (permissive licence) · get_code("30b05f2a415dd132")
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 OrderedDict. 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 collections
import six
import tempfile
import os

# Create a temporary vocabulary file
temp_dir = tempfile.mkdtemp()
vocab_file = os.path.join(temp_dir, "vocab.txt")

# Write a small vocabulary file
vocab_content = "hello\nworld\n[CLS]\n[SEP]\n[UNK]\n"
with open(vocab_file, 'w') as f:
    f.write(vocab_content)

ARGS = {
    'vocab_file': vocab_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
six, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type OrderedDict — a OrderedDict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a OrderedDict 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
load_vocab Ran
ran; our contract draft was wrong, not the code
coronazap/bert_client/tokenization.py
code served (permissive licence) · get_code("21c6f1a7ac31a2bd")
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 OrderedDict. 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 collections
import six
import tempfile
import os

# Create a temporary vocab file for testing
temp_dir = tempfile.mkdtemp()
vocab_file = os.path.join(temp_dir, "vocab.txt")

# Write a small vocab file
vocab_content = ["hello", "world", "test", "python", "code"]
with open(vocab_file, "w") as f:
    for word in vocab_content:
        f.write(word + "\n")

ARGS = {
    "vocab_file": vocab_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
six, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type OrderedDict — a OrderedDict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a OrderedDict 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_qid_to_has_ans Ran
ran; our contract draft was wrong, not the code
somiltg/bert/evaluate-v2.0.py
code served (permissive licence) · get_code("e91c44f5680827e2")
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 dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Construct a minimal dataset matching the expected structure
dataset = [
    {
        'paragraphs': [
            {
                'qas': [
                    {
                        'id': 'q1',
                        'answers': [{'text': 'answer1', 'answer_start': 0}]
                    },
                    {
                        'id': 'q2',
                        'answers': []
                    }
                ]
            },
            {
                'qas': [
                    {
                        'id': 'q3',
                        'answers': [{'text': 'answer3', 'answer_start': 5}]
                    }
                ]
            }
        ]
    },
    {
        'paragraphs': [
            {
                'qas': [
                    {
                        'id': 'q4',
                        'answers': []
                    }
                ]
            }
        ]
    }
]

ARGS = {
    'dataset': dataset
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
normalize_answer Ran
ran; our contract draft was wrong, not the code
somiltg/bert/evaluate-v2.0.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from saptarshi059/cdqa-project, is pointer only (licence: NONE) · get_code("6a96435eba311b08")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from saptarshi059/cdqa-project, and GitHub detected no licence in that repository 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.

printable_text Ran
ran; our contract draft was wrong, not the code
Arthurizijar/Bert_Airport/tokenization.py
code served (permissive licence) · get_code("0e5615f8994003cf")
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 six

# The function takes a single parameter 'text' which can be:
# - str (works in both Python 2 and 3)
# - bytes (in Python 3, decoded to utf-8)
# - unicode (in Python 2, encoded to utf-8)

# Since we're running on Python 3, we'll test with a string
text_input = "Hello, World!"

ARGS = {
    "text": text_input
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
six — 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
process_result Ran
ran on a synthesized input
coronazap/bert_client/run_squad.py
code served (permissive licence) · get_code("b57e2fc59104beb6")
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 RawResult. 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 collections

RawResult = collections.namedtuple('RawResult', ['unique_id', 'start_logits', 'end_logits'])

class MockTensorValue:
    def __init__(self, value_list, is_int=False):
        if is_int:
            self.int64_val = value_list
            self.float_val = []
        else:
            self.float_val = value_list
            self.int64_val = []

result = {
    'unique_ids': MockTensorValue([42], is_int=True),
    'start_logits': MockTensorValue([0.1, 0.2, 0.3, 0.4]),
    'end_logits': MockTensorValue([0.5, 0.6, 0.7, 0.8])
}

ARGS = {
    'result': result
}
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 RawResult — a RawResult 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
set_logger Ran
ran; our contract draft was wrong, not the code
StoneGH/bert/freeze_graph.py
code served (permissive licence) · get_code("9b7c0abd85734790")
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 Logger. 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
import logging

torch_imported = False
try:
    import torch
    torch_imported = True
except ImportError:
    pass

# Construct arguments for set_logger function
ARGS = {
    'context': 'test_logger',
    'verbose': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Logger — a Logger output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Logger 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
clones Not yet run
unverified (harvested, no recorded run)
the-Quert/iNLPfun/custom/encoder_ref.py
code served (permissive licence) · get_code("7ba03346caea013d")
How it ran

no invocation recorded.

create_optimizer Not yet run
unverified (harvested, no recorded run)
MOHAMEDELDAKDOUKY/bert_adjusted/optimization.py
code served (permissive licence) · get_code("c070006ff55cd83a")
How it ran

no invocation recorded.

download_and_load_datasets Not yet run
unverified (harvested, no recorded run)
Chonwai/Learning_BERT/predicting_movie_reviews_with_bert_on_tf_hub.py
code served (permissive licence) · get_code("85813dd1e897d90f")
How it ran

no invocation recorded.

file_based_input_fn_builder Not yet run
unverified (harvested, no recorded run)
Arthurizijar/Bert_Airport/run_classifier.py
code served (permissive licence) · get_code("1b46c9ce24fc6db3")
How it ran

no invocation recorded.

file_based_input_fn_builder Not yet run
unverified (harvested, no recorded run)
Maz101/Bert/run_classifier.py
code served (permissive licence) · get_code("28b353970e44ff54")
How it ran

no invocation recorded.

get_assignment_map_from_checkpoint Not yet run
unverified (harvested, no recorded run)
Arthurizijar/Bert_Airport/modeling.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from toggle1995/RIS-DMMI, is pointer only (licence: NONE) · get_code("50958618b65e514e")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from toggle1995/RIS-DMMI, and GitHub detected no licence in that repository 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. No invocation recorded.

load_dataset Not yet run
unverified (harvested, no recorded run)
Chonwai/Learning_BERT/predicting_movie_reviews_with_bert_on_tf_hub.py
code served (permissive licence) · get_code("a181d94c3334306d")
How it ran

no invocation recorded.

load_directory_data Not yet run
unverified (harvested, no recorded run)
Chonwai/Learning_BERT/predicting_movie_reviews_with_bert_on_tf_hub.py
code served (permissive licence) · get_code("0ccbf8e553bab00d")
How it ran

no invocation recorded.

load_vocab Not yet run
unverified (harvested, no recorded run)
Arthurizijar/Bert_Airport/tokenization.py
code served (permissive licence) · get_code("ff83ccc8b0b6462d")
How it ran

no invocation recorded.

process_output Not yet run
unverified (harvested, no recorded run)
coronazap/bert_client/run_squad.py
code served (permissive licence) · get_code("7b79d58035cd12c2")
How it ran

no invocation recorded.

read_squad_examples Not yet run
unverified (harvested, no recorded run)
coronazap/bert_client/run_squad.py
code served (permissive licence) · get_code("3cbc0c492348e83d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 32 samples lifted from this paper's linked repositories, 32 carry a function identity and 0 do not. Of the 30 distinct functions with an identity, 19 are also present in other papers whose code we hold (17 with at least one paper dated earlier, 1 the same month, 18 later) and 11 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
normalize_answer 37 other papers
3 dated earlier, 1 the same month, 33 later · 1 only as an assembly brick
1808.07042 (2018-08)
CoQA: A Conversational Question Answering Challenge
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1902.01718 (2019-02)
End-to-End Open-Domain Question Answering with BERTserini
and 34 more not named here
convert_to_unicode 36 other papers
9 dated earlier, 27 later
1609.08144 (2016-09)
Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1901.04085 (2019-01)
Passage Re-ranking with BERT
and 33 more not named here
printable_text 33 other papers
9 dated earlier, 24 later
1609.08144 (2016-09)
Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1901.04085 (2019-01)
Passage Re-ranking with BERT
and 30 more not named here
get_activation 32 other papers
10 dated earlier, 22 later · 1 only as an assembly brick
1804.05392 (2018-04)
Higher-order Coreference Resolution with Coarse-to-fine Inference
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1901.04085 (2019-01)
Passage Re-ranking with BERT
and 29 more not named here
gelu 26 other papers
7 dated earlier, 19 later · 4 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1901.02860 (2019-01)
Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context
and 23 more not named here
get_assignment_map_from_checkpoint 26 other papers
10 dated earlier, 16 later
1804.05392 (2018-04)
Higher-order Coreference Resolution with Coarse-to-fine Inference
1808.09781 (2018-08)
Self-Attentive Sequential Recommendation
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
and 23 more not named here
get_tokens 25 other papers
2 dated earlier, 23 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1902.01718 (2019-02)
End-to-End Open-Domain Question Answering with BERTserini
1910.01244 (2019-10)
Linking artificial and human neural representations of language
and 22 more not named here
load_vocab 16 other papers
5 dated earlier, 11 later
1609.08144 (2016-09)
Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1901.08746 (2019-01)
BioBERT: a pre-trained biomedical language representation model for biomedical text mining
and 13 more not named here
create_int_feature 13 other papers
2 dated earlier, 11 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1906.05474 (2019-06)
Transfer Learning in Biomedical Natural Language Processing: An Evaluation of BERT and ELMo on Ten Benchmarking Datasets
1909.06356 (2019-09)
Addressing Semantic Drift in Question Generation for Semi-Supervised Question Answering
and 10 more not named here
create_float_feature 9 other papers
1 dated earlier, 8 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1909.06356 (2019-09)
Addressing Semantic Drift in Question Generation for Semi-Supervised Question Answering
1909.11942 (2019-09)
ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
and 6 more not named here
make_qid_to_has_ans 9 other papers
1 dated earlier, 8 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1910.01244 (2019-10)
Linking artificial and human neural representations of language
2004.13922 (2020-04)
Revisiting Pre-trained Models for Chinese Natural Language Processing
and 6 more not named here
file_based_input_fn_builder 7 other papers
3 dated earlier, 4 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1905.09217 (2019-05)
Deeper Text Understanding for IR with Contextual Neural Language Modeling
1906.09317 (2019-06)
Identification of Tasks, Datasets, Evaluation Metrics, and Numeric Scores for Scientific Leaderboards Construction
and 4 more not named here
input_fn_builder 6 other papers
1 dated earlier, 5 later
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
1909.06356 (2019-09)
Addressing Semantic Drift in Question Generation for Semi-Supervised Question Answering
1910.04732 (2019-10)
Structured Pruning of Large Language Models
and 3 more not named here
load_dataset 3 other papers
2 dated earlier, 1 later
1606.01305 (2016-06)
Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
2105.00303 (2021-05)
RATT: Leveraging Unlabeled Data to Guarantee Generalization
load_directory_data 3 other papers
2 dated earlier, 1 later
1606.01305 (2016-06)
Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
2105.00303 (2021-05)
RATT: Leveraging Unlabeled Data to Guarantee Generalization
download_and_load_datasets 2 other papers
1 dated earlier, 1 later
1606.01305 (2016-06)
Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations
2105.00303 (2021-05)
RATT: Leveraging Unlabeled Data to Guarantee Generalization
load_vocab 2 other papers
2 later
1909.11942 (2019-09)
ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
2006.01563 (2020-06)
Exploring Cross-sentence Contexts for Named Entity Recognition with BERT
create_optimizer 1 other paper
1 dated earlier
1810.04805 (2018-10)
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
set_logger 1 other paper
1 later
2205.13190 (2022-05)
Other Roles Matter! Enhancing Role-Oriented Dialogue Summarization via Role Interactions

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 2026-11-21.

Repositories linked to this paper

More are linked than shown here.

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

Abstract

Recent developments in natural language representations have been accompanied by large and expensive models that leverage vast amounts of general-domain text through self-supervised pre-training. Due to the cost of applying such models to down-stream tasks, several model compression techniques on pre-trained language representations have been proposed (Sun et al., 2019; Sanh, 2019). However, surprisingly, the simple baseline of just pre-training and fine-tuning compact models has been overlooked. In this paper, we first show that pre-training remains important in the context of smaller architectures, and fine-tuning pre-trained compact models can be competitive to more elaborate methods proposed in concurrent work. Starting with pre-trained compact models, we then explore transferring task knowledge from large fine-tuned models through standard knowledge distillation. The resulting simple, yet effective and general algorithm, Pre-trained Distillation, brings further improvements. Through extensive experiments, we more generally explore the interaction between pre-training and distillation under two variables that have been under-studied: model size and properties of unlabeled task data. One surprising observation is that they have a compound effect even when sequentially applied on the same data. To accelerate future research, we will make our 24 pre-trained miniature BERT models publicly available.

For agents

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

get_harvested_code_for_paper("1908.08962")
get_code_for_paper("1908.08962")
have("1908.08962")

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

curl https://syntology.ai/api/ran/1908.08962.json

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

[![Syntology run record](https://syntology.ai/api/ran/1908.08962.svg)](https://syntology.ai/paper/1908.08962)

Connect an agent — have() is free.