SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1909.09436 · 2019

CodeSearchNet Challenge: Evaluating the State of Semantic Code Search

arXiv · PDF · Open in the Atlas

Code that ran

15 of 17 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 7 have no contract verdict; 7 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/1909.09436.json · badge: /api/ran/1909.09436.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
github/CodeSearchNet canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
nchen909/codeattention community (classifier label)
classifier basis not recorded on the link
8 ran · 2 no recorded run
jmg764/CodeSearchNet-Challenge community (classifier label)
classifier basis not recorded on the link
4 ran · 0 no recorded run
FunctionStatusWhere it lives
add_lang_by_task Ran
ran; our contract draft was wrong, not the code
nchen909/codeattention/utils.py
code served (permissive licence) · get_code("2e3e72e1e860afb2")
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 torch

ARGS = {
    'target_str': 'This is a test string',
    'task': 'summarize',
    'sub_task': 'java-cs'
}
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 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
compute_bleu Ran
ran; our contract draft was wrong, not the code
nchen909/codeattention/evaluator/bleu.py
pointer only (licence: MIT); the stored copy get_code serves, from WangXFng/RDRec, is pointer only (licence: NONE) · get_code("49289a7f94c227bd")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from WangXFng/RDRec, 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.

convert_clone_examples_to_features Ran
ran on a synthesized input
nchen909/codeattention/utils.py
code served (permissive licence) · get_code("cfd0293d796a30d2")
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 CloneInputFeatures. 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 encode(self, text, max_length=None, padding=None, truncation=None):
        # Simple tokenizer that returns a list of token ids
        tokens = [101] + [i % 1000 for i in range(min(len(text), max_length - 2 if max_length else len(text)))] + [102]
        if max_length:
            if len(tokens) < max_length:
                tokens = tokens + [0] * (max_length - len(tokens))
            else:
                tokens = tokens[:max_length]
        return tokens

class SimpleExample:
    def __init__(self, source, target, label, url1, url2):
        self.source = source
        self.target = target
        self.label = label
        self.url1 = url1
        self.url2 = url2

class SimpleArgs:
    def __init__(self):
        self.model_name = 'bert'
        self.add_task_prefix = False
        self.task = 'clone_detection'
        self.block_size = 8

# Create test data
example = SimpleExample(
    source="def foo(): return 42",
    target="def bar(): return 42",
    label=1,
    url1="http://example.com/1",
    url2="http://example.com/2"
)

example_index = 0
tokenizer = SimpleTokenizer()
args = SimpleArgs()

ARGS = {
    'item': (example, example_index, tokenizer, args)
}
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 CloneInputFeatures — a CloneInputFeatures output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a CloneInputFeatures 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_examples_to_features Ran
ran on a synthesized input
nchen909/codeattention/utils.py
code served (permissive licence) · get_code("fa57c04c2fcf72af")
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
from transformers import AutoTokenizer

class MockExample:
    def __init__(self, source, target, url=None):
        self.source = source
        self.target = target
        self.url = url

class MockTokenizer:
    def __init__(self):
        self.eos_token_id = 2
    
    def encode(self, text, max_length=None, padding=None, truncation=None):
        # Return a list with exactly one eos_token_id at the end
        tokens = [1] * (max_length - 1) if max_length else [1] * 9
        tokens.append(self.eos_token_id)
        return tokens

class MockArgs:
    def __init__(self):
        self.model_name = 't5'
        self.add_task_prefix = True
        self.task = 'summarize'
        self.sub_task = 'none'
        self.add_lang_ids = True
        self.max_source_length = 10
        self.max_target_length = 10

example = MockExample(
    source="def hello(): return 42",
    target="function that returns 42",
    url="http://example.com"
)

tokenizer = MockTokenizer()

args = MockArgs()

ARGS = {
    'item': (
        example,
        0,
        tokenizer,
        args,
        'train'
    )
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — 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
count_ngrams Ran
ran; our contract draft was wrong, not the code
nchen909/codeattention/evaluator/smooth_bleu.py
code served (permissive licence) · get_code("835fca2953749b59")
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

torch.manual_seed(0)

# The function count_ngrams takes:
# - words: a sequence of items (strings, tokens, etc.)
# - n: an integer specifying the maximum n-gram size

# Create a simple list of words
words = ["the", "quick", "brown", "fox", "jumps", "over", "the", "lazy", "dog"]

# Create the ARGS dictionary with all parameters
ARGS = {
    "words": words,
    "n": 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
coverage_per_language Ran
ran, honoured the contract we drafted
github/CodeSearchNet/src/relevanceeval.py
code served (permissive licence) · get_code("b9d46fdb66261637")
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 float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Construct valid arguments for coverage_per_language

# predictions: Dict[str, List[str]]
# Maps query strings to lists of predicted URLs
predictions = {
    "query1": ["url_a", "url_b", "url_c"],
    "query2": ["url_x", "url_y"],
    "query3": ["url_p", "url_q", "url_r", "url_s"]
}

# relevance_scores: Dict[str, Dict[str, float]]
# Maps query strings to dicts of URL -> relevance score
relevance_scores = {
    "query1": {
        "url_a": 1.0,
        "url_b": 0.5,
        "url_d": -0.2,
        "url_e": 0.8
    },
    "query2": {
        "url_x": 0.9,
        "url_z": -0.1,
        "url_y": 0.3
    },
    "query3": {
        "url_p": 1.0,
        "url_q": 0.0,
        "url_r": 0.7,
        "url_t": -0.5
    }
}

# with_positive_relevance: bool
with_positive_relevance = False

ARGS = {
    "predictions": predictions,
    "relevance_scores": relevance_scores,
    "with_positive_relevance": with_positive_relevance
}
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 float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_model_size Ran
ran; our contract draft was wrong, not the code
nchen909/codeattention/models.py
code served (permissive licence) · get_code("8cfe38019e9d6a49")
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 torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

# Create a minimal model with some trainable parameters
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

ARGS = {
    'model': model
}
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 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
get_token_attention Ran
ran on a synthesized input
nchen909/codeattention/attention.py
code served (permissive licence) · get_code("7b76b21df15d335b")
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 a temporary directory for cache
temp_dir = tempfile.mkdtemp()

# Create a simple args class
class Args:
    def __init__(self, cache_path, pickle_suffix):
        self.cache_path = cache_path
        self.pickle_suffix = pickle_suffix

# Construct the arguments
args = Args(cache_path=temp_dir, pickle_suffix='_test')

# attention_numpy: shape (batch_size, num_layers, num_heads, seq_len, seq_len)
# We need shape (2, 1, 3, 5, 5) - batch=2, layers=1, heads=3, seq=5, seq=5
attention_numpy = np.random.rand(2, 1, 3, 5, 5).astype(np.float32)

# subtoken_numbers_list: list of lists mapping subtokens to token indices
# For batch size 2, each element maps 5 subtokens to token numbers
# Using -1 to indicate padding/skip, and valid token indices
subtoken_numbers_list = [
    [0, 0, 1, 1, 2],  # batch 0: subtokens map to tokens 0, 0, 1, 1, 2
    [0, 1, 1, 2, -1]  # batch 1: subtokens map to tokens 0, 1, 1, 2, -1 (skip)
]

# action_per_head: either 'mean' or 'max'
action_per_head = 'mean'

ARGS = {
    'args': args,
    'attention_numpy': attention_numpy,
    'subtoken_numbers_list': subtoken_numbers_list,
    'action_per_head': action_per_head
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tqdm — 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
label_folds Ran
ran on a synthesized input
jmg764/CodeSearchNet-Challenge/src/dataextraction/dedup_split.py
code served (permissive licence) · get_code("3db12c135869bd1e")
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 DataFrame. 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 pandas as pd

# Create a small DataFrame with the required columns: 'repo' and 'path'
df = pd.DataFrame({
    'repo': ['repo_a', 'repo_b', 'repo_c', 'repo_d', 'repo_e'],
    'path': ['path/to/file1', 'path/to/file2', 'path/to/file3', 'path/to/file4', 'path/to/file5']
})

ARGS = {
    'df': df,
    'train_ratio': 0.6,
    'valid_ratio': 0.2,
    'test_ratio': 0.1,
    'holdout_ratio': 0.1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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_predictions Ran
ran; our contract draft was wrong, not the code
github/CodeSearchNet/src/relevanceeval.py
code served (permissive licence) · get_code("faa62c1fbe7df43e")
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 defaultdict. 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 pandas as pd
import tempfile
import os

# Create a temporary CSV file with sample prediction data
temp_dir = tempfile.mkdtemp()
filepath = os.path.join(temp_dir, "predictions.csv")

# Create sample data
data = {
    'language': ['English', 'english', 'French', 'ENGLISH', 'Spanish'],
    'query': ['machine learning', 'Machine Learning', 'deep learning', 'neural networks', 'data science'],
    'url': [
        'https://example.com/ml1',
        'https://example.com/ml2',
        'https://example.com/dl1',
        'https://example.com/nn1',
        'https://example.com/ds1'
    ]
}

df = pd.DataFrame(data)
df.to_csv(filepath, index=False)

ARGS = {
    'filepath': filepath,
    'max_urls_per_language': 300
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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_relevances Ran
ran; our contract draft was wrong, not the code
github/CodeSearchNet/src/relevanceeval.py
code served (permissive licence) · get_code("9279f17104890350")
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 defaultdict. 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 pandas as pd
import numpy as np

# Create a temporary CSV file with sample relevance data
temp_file = tempfile.NamedTemporaryFile(mode='w', suffix='.csv', delete=False)
temp_filepath = temp_file.name

# Write sample data to the CSV
csv_data = """Query,Language,GitHubUrl,Relevance
python sorting,Python,https://github.com/user/repo1,0.9
python sorting,Python,https://github.com/user/repo2,0.8
python sorting,JavaScript,https://github.com/user/repo3,0.5
web framework,Python,https://github.com/user/repo4,0.7
web framework,JavaScript,https://github.com/user/repo5,0.85
web framework,JavaScript,https://github.com/user/repo6,0.75
"""

temp_file.write(csv_data)
temp_file.close()

ARGS = {
    'filepath': temp_filepath
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type defaultdict — a defaultdict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a defaultdict 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
pad_sequence Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
nchen909/codeattention/evaluator/CodeBLEU/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from alex-haochenli/reco, is pointer only (licence: NONE) · get_code("cd90fe437a24c233")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from alex-haochenli/reco, 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.

sample_data Ran
ran on a synthesized input
jmg764/CodeSearchNet-Challenge/src/error_analysis.py
code served (permissive licence) · get_code("4071a311ada99ce3")
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
from random import sample
from typing import List, Dict, Any, Optional

torch.manual_seed(0)

# Create sample data with distinct entries
data = [
    {"id": 1, "text": "example one", "value": 10},
    {"id": 2, "text": "example two", "value": 20},
    {"id": 3, "text": "example three", "value": 30},
    {"id": 4, "text": "example four", "value": 40},
    {"id": 5, "text": "example five", "value": 50},
]

ARGS = {
    "data": data,
    "max_num_examples": 3,
}
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
to_highlighted_html Ran
ran on a synthesized input
jmg764/CodeSearchNet-Challenge/src/error_analysis.py
code served (permissive licence) · get_code("9dbbcd2b21d49a73")
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.

from pygments import highlight
from pygments.lexers import get_lexer_by_name
from pygments.formatters import HtmlFormatter

def to_highlighted_html(code: str, language: str) -> str:
    lexer = get_lexer_by_name(language, stripall=True)
    formatter = HtmlFormatter(linenos=True)
    return highlight(code, lexer, formatter)


ARGS = {
    "code": "def hello():\n    print('world')",
    "language": "python"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pygments — 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
tokenize_docstring_from_string Ran
ran on a synthesized input
jmg764/CodeSearchNet-Challenge/src/dataextraction/utils.py
code served (permissive licence) · get_code("9140bccc6c9f6468")
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 re
from typing import List

DOCSTRING_REGEX_TOKENIZER = re.compile('[^\\s,\'\\"`.():\\[\\]=*;>{\\}+-/\\\\]+|\\\\+|\\.+|\\(\\)|{\\}|\\[\\]|\\(+|\\)+|:+|\\[+|\\]+|{+|\\}+|=+|\\*+|;+|>+|\\++|-+|/+')

def tokenize_docstring_from_string(docstr: str) -> List[str]:
    return [t for t in DOCSTRING_REGEX_TOKENIZER.findall(docstr) if t is not None and len(t) > 0]

# Create a sample docstring to tokenize
sample_docstring = """
This is a sample function docstring.
It includes various tokens: function(), [brackets], {braces}, and special chars like * + - / and \\ backslash.
Parameters: x, y, z
Returns: result
Example: foo(bar=123)
"""

ARGS = {
    "docstr": sample_docstring
}
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
add_args Not yet run
unverified (harvested, no recorded run)
nchen909/codeattention/configs.py
code served (permissive licence) · get_code("91fcc17204abe9ed")
How it ran

no invocation recorded.

bulid_or_load_gen_model Not yet run
unverified (harvested, no recorded run)
nchen909/codeattention/models.py
code served (permissive licence) · get_code("62f9216d4bcf84c0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 17 distinct functions with an identity, 5 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 5 later) and 12 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
compute_bleu 10 other papers
10 later
aaai_6453 (2020 (year only))
Improving Knowledge-Aware Dialogue Generation via Knowledge Base Question Answering
2206.05239 (2022-06)
StructCoder: Structure-Aware Transformer for Code Generation
2206.08474 (2022-06)
XLCoST: A Benchmark Dataset for Cross-lingual Code Intelligence
and 7 more not named here
pad_sequence 10 other papers
10 later
1911.04700 (2019-11)
A Pre-training Based Personalized Dialogue Generation Model with Persona-sparse Data
2009.10297 (2020-09)
CodeBLEU: a Method for Automatic Evaluation of Code Synthesis
2106.06169 (2021-06)
BoB: BERT Over BERT for Training Persona-based Dialogue Models from Limited Personalized Data
and 7 more not named here
get_model_size 6 other papers
6 later
2206.07796 (2022-06)
FixEval: Execution-based Evaluation of Program Fixes for Programming Problems
2207.11680 (2022-07)
No More Fine-Tuning? An Experimental Evaluation of Prompt Tuning in Code Intelligence
2305.17506 (2023-05)
Backdooring Neural Code Search
and 3 more not named here
add_lang_by_task 4 other papers
4 later
2206.07796 (2022-06)
FixEval: Execution-based Evaluation of Program Fixes for Programming Problems
2207.11680 (2022-07)
No More Fine-Tuning? An Experimental Evaluation of Prompt Tuning in Code Intelligence
2311.00317 (2023-11)
Data Augmentation for Code Translation with Comparable Corpora and Multiple References
and 1 more not named here
count_ngrams 3 other papers
3 later
1910.07517 (2019-10)
Adversarial Examples for Models of Code
2109.08365 (2021-09)
CodeQA: A Question Answering Dataset for Source Code Comprehension
2207.11680 (2022-07)
No More Fine-Tuning? An Experimental Evaluation of Prompt Tuning in Code Intelligence

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

Semantic code search is the task of retrieving relevant code given a natural language query. While related to other information retrieval tasks, it requires bridging the gap between the language used in code (often abbreviated and highly technical) and natural language more suitable to describe vague concepts and ideas. To enable evaluation of progress on code search, we are releasing the CodeSearchNet Corpus and are presenting the CodeSearchNet Challenge, which consists of 99 natural language queries with about 4k expert relevance annotations of likely results from CodeSearchNet Corpus. The corpus contains about 6 million functions from open-source code spanning six programming languages (Go, Java, JavaScript, PHP, Python, and Ruby). The CodeSearchNet Corpus also contains automatically generated query-like natural language for 2 million functions, obtained from mechanically scraping and preprocessing associated function documentation. In this article, we describe the methodology used to obtain the corpus and expert labels, as well as a number of simple baseline solutions for the task. We hope that CodeSearchNet Challenge encourages researchers and practitioners to study this interesting task further and will host a competition and leaderboard to track the progress on the challenge. We are also keen on extending CodeSearchNet Challenge to more queries and programming languages in the future.

For agents

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

get_harvested_code_for_paper("1909.09436")
get_code_for_paper("1909.09436")
have("1909.09436")

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

curl https://syntology.ai/api/ran/1909.09436.json

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

[![Syntology run record](https://syntology.ai/api/ran/1909.09436.svg)](https://syntology.ai/paper/1909.09436)

Connect an agent — have() is free.