SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1611.09268 · 2016

MS MARCO: A Human Generated MAchine Reading COmprehension Dataset

arXiv · PDF · Open in the Atlas

Code that ran

28 of 33 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; 0 honoured the contract we drafted, 0 violated it and 23 have no contract verdict; 5 failures of our instrument; 5 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/1611.09268.json · badge: /api/ran/1611.09268.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
microsoft/OpenKP canonical (classifier label)
classifier basis: census
5 ran · 1 no recorded run
microsoft/MSMARCO-Passage-Ranking canonical (classifier label)
classifier basis: census
5 ran · 1 no recorded run
dfcf93/MSMARCO community (classifier label)
classifier basis not recorded on the link
8 ran · 1 no recorded run
microsoft/MSMARCO-Question-Answering community (classifier label)
classifier basis not recorded on the link
6 ran · 0 no recorded run
AmenRa/rank_eval community (classifier label)
classifier basis not recorded on the link
3 ran · 2 no recorded run
amenra/ranx community (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
FunctionStatusWhere it lives
cook_refs Ran
ran; our contract draft was wrong, not the code
microsoft/MSMARCO-Question-Answering/Evaluation/bleu_scorer.py
code served (permissive licence) · get_code("7586e1d32f7325f7")
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.

from collections import defaultdict

ARGS = {
    'refs': [
        'the quick brown fox jumps over the lazy dog',
        'a quick brown fox leaps over a lazy dog'
    ],
    'eff': 'shortest',
    'n': 4
}
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
cook_test Ran
ran; our contract draft was wrong, not the code
microsoft/MSMARCO-Question-Answering/Evaluation/bleu_scorer.py
code served (permissive licence) · get_code("019ae8bc6f98133f")
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.

from collections import defaultdict

def precook(s, n=4, out=False):
    """Takes a string as input and returns an object that can be given to
    either cook_refs or cook_test. This is optional: cook_refs and cook_test
    can take string arguments as well."""
    words = s.split()
    counts = defaultdict(int)
    for k in range(1, n + 1):
        for i in range(len(words) - k + 1):
            ngram = tuple(words[i:i + k])
            counts[ngram] += 1
    return (len(words), counts)

def cook_test(test, xxx_todo_changeme, eff=None, n=4):
    """Takes a test sentence and returns an object that
    encapsulates everything that BLEU needs to know about it."""
    reflen, refmaxcounts = xxx_todo_changeme
    testlen, counts = precook(test, n, True)
    result = {}
    if eff == 'closest':
        result['reflen'] = min(((abs(l - testlen), l) for l in reflen))[1]
    else:
        result['reflen'] = reflen
    result['testlen'] = testlen
    result['guess'] = [max(0, testlen - k + 1) for k in range(1, n + 1)]
    result['correct'] = [0] * n
    for ngram, count in counts.items():
        result['correct'][len(ngram) - 1] += min(refmaxcounts.get(ngram, 0), count)
    return result

# Construct test string
test_string = "the quick brown fox jumps over the lazy dog"

# Construct reference data (precook output)
ref_text = "the quick brown fox jumps over a lazy dog"
ref_length, ref_counts = precook(ref_text, n=4)

# Build xxx_todo_changeme parameter
# It expects (reflen, refmaxcounts)
# reflen should be iterable of lengths, refmaxcounts should be a defaultdict
xxx_todo_changeme_value = ([ref_length], ref_counts)

ARGS = {
    'test': test_string,
    'xxx_todo_changeme': xxx_todo_changeme_value,
    'eff': None,
    'n': 4
}
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
dedup Ran
ran on a synthesized input
microsoft/OpenKP/evaluate.py
code served (permissive licence) · get_code("2a9560b1d7891c67")
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 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function dedup takes a kp_list and removes duplicates while preserving order.
# It checks if each kp is hashable and can be added to a set.
# Let's create a list of hashable keypoints (tuples of floats work well as they're hashable)

kp_list = [
    (1.0, 2.0, 3.0, 4.0),
    (5.0, 6.0, 7.0, 8.0),
    (1.0, 2.0, 3.0, 4.0),  # duplicate
    (9.0, 10.0, 11.0, 12.0),
    (5.0, 6.0, 7.0, 8.0),  # duplicate
]

ARGS = {
    "kp_list": kp_list
}
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 2, 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
generateOuput Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateQueryEmbeddings.py
code served (permissive licence) · get_code("537a8f2a5574b5d6")
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 numpy as np

torch.manual_seed(0)

# Construct responses list with required structure
responses = [
    {
        'query': 'what is python',
        'vector': [0.1, 0.2, 0.3, 0.4]
    },
    {
        'query': 'machine learning basics',
        'vector': [0.5, 0.6, 0.7, 0.8]
    }
]

ARGS = {
    'responses': responses
}
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
getScoreEM Ran
ran on a synthesized input
microsoft/OpenKP/agreement.py
code served (permissive licence) · get_code("5bbdf43246de0b42")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Analyzing the function:
# - candidate: iterable of labels that can be converted to strings and compared with ==
# - gold: iterable of labels that can be converted to strings and compared with ==
# - The function compares candidate and gold labels for equality
# - It returns a score as sum(bestMatch.values()) / len(gold)
# - Both parameters are used in for loops and string conversion

# Create simple label candidates and gold standard
candidate = ['label_A', 'label_B', 'label_C']
gold = ['label_A', 'label_B']

ARGS = {
    'candidate': candidate,
    'gold': gold
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
getScoreUnigram Ran
ran on a synthesized input
microsoft/OpenKP/agreement.py
code served (permissive licence) · get_code("1789c669f2709136")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Analyzing the function:
# - candidate and gold are iterable collections of items
# - The function performs set operations on these items (using - operator)
# - It calculates len() on items and uses len() on the collections
# - Items need to support subtraction (set-like behavior)

# Using frozenset as the item type since it supports:
# - len()
# - set difference operator (-)
# - hashing (needed for dict keys via str())

candidate = [
    frozenset([1, 2, 3]),
    frozenset([2, 3, 4]),
]

gold = [
    frozenset([1, 2]),
    frozenset([2, 3, 4, 5]),
]

ARGS = {
    "candidate": candidate,
    "gold": gold,
}
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 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_candidates Ran
ran on a synthesized input
amenra/ranx/ranx/fusion/bordafuse.py
code served (permissive licence) · get_code("ccb4d73aaca976aa")
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. 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
from numba.typed import List as TypedList
from numba.types import unicode_type

# Construct runs parameter
# runs is expected to be a list of dictionaries
# Each dictionary maps query_ids (keys) to nested dictionaries of doc_ids

run1 = {
    "q1": {"doc_1": 0.9, "doc_2": 0.8, "doc_3": 0.7},
    "q2": {"doc_4": 0.95, "doc_5": 0.85},
    "q3": {"doc_1": 0.75, "doc_6": 0.65}
}

run2 = {
    "q1": {"doc_2": 0.88, "doc_7": 0.72},
    "q2": {"doc_4": 0.92, "doc_8": 0.80},
    "q3": {"doc_9": 0.78}
}

runs = [run1, run2]

ARGS = {
    "runs": runs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type List — a List 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
get_file_kind Ran
ran on a synthesized input
AmenRa/rank_eval/ranx/data_structures/qrels.py
code served (permissive licence) · get_code("77f6de2e73ca8660")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import os

def get_file_kind(path: str='qrels.json', kind: str=None) -> str:
    if kind is None:
        kind = os.path.splitext(path)[1][1:]
        kind = 'trec' if kind == 'txt' else kind
        kind = 'parquet' if kind == 'parq' else kind
    assert kind in {'json', 'trec', 'gz', 'parquet'}, "Error `kind` must be 'json', 'trec', 'gz', or 'parquet'"
    return kind

ARGS = {
    'path': 'qrels.json',
    'kind': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_metric_label Ran
ran on a synthesized input
AmenRa/rank_eval/ranx/ranxhub.py
code served (permissive licence) · get_code("ec8c6814aa969fd2")
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

# The function takes a single parameter 'm' which is a string metric name
# The function checks if '-l' is in m, splits on '@', and looks up the metric in metric_labels dict

ARGS = {
    'm': 'ndcg@10'
}
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
get_stats Ran
ran on a synthesized input
microsoft/MSMARCO-Question-Answering/Utils/get_stats_about_length.py
code served (permissive licence) · get_code("9346371b965cc92e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# The function expects a histogram parameter which is a dict-like object
# where keys are values and dict[value] gives the frequency/count
# The function iterates over keys and accesses histogram[value]

histogram = {
    1.0: 5,      # value 1.0 appears 5 times
    2.0: 3,      # value 2.0 appears 3 times
    3.0: 2,      # value 3.0 appears 2 times
}

ARGS = {
    "histogram": histogram
}
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 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_url Ran
ran on a synthesized input
AmenRa/rank_eval/ranx/io.py
code served (permissive licence) · get_code("76ec290b848f6d2d")
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

# Analyze the function:
# - get_url takes a single parameter 'id' of type str
# - It uses the id in an f-string with quote() to construct a URL
# - The function simply returns a formatted string

ARGS = {
    'id': 'my_test_id_123'
}
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
loadMSMARCO Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateQuerysets.py
code served (permissive licence) · get_code("19ad04a1d7cd706f")
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 set. 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 temporary files with MSMARCO-like format
# The function expects files where each line has tab-separated values
# and the second column (index 1) contains query text

temp_dir = tempfile.mkdtemp()

# Create first temporary file
file1 = os.path.join(temp_dir, "file1.tsv")
with open(file1, 'w') as f:
    f.write("doc_id_1\tquery one\tpassage one\n")
    f.write("doc_id_2\tquery two\tpassage two\n")
    f.write("doc_id_3\tquery one\tpassage three\n")

# Create second temporary file
file2 = os.path.join(temp_dir, "file2.tsv")
with open(file2, 'w') as f:
    f.write("doc_id_4\tquery three\tpassage four\n")
    f.write("doc_id_5\tquery two\tpassage five\n")

ARGS = {
    "files": [file1, file2]
}
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 set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
loadNQ Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateQuerysets.py
code served (permissive licence) · get_code("f35b559c8eea08c7")
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 set. 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 json
import os

# Create a temporary directory with test files
temp_dir = tempfile.mkdtemp()

# Create test files with NQ-format data
test_data_1 = [
    {'question_text': 'What is the capital of France?'},
    {'question_text': 'Who wrote Romeo and Juliet?'},
    {'question_text': 'What is the capital of France?'},  # duplicate
]

test_data_2 = [
    {'question_text': 'What is the largest planet?'},
    {'question_text': 'Who wrote Romeo and Juliet?'},  # duplicate from file 1
]

# Write test files
with open(os.path.join(temp_dir, 'file1.jsonl'), 'w') as f:
    for item in test_data_1:
        f.write(json.dumps(item) + '\n')

with open(os.path.join(temp_dir, 'file2.jsonl'), 'w') as f:
    for item in test_data_2:
        f.write(json.dumps(item) + '\n')

ARGS = {
    'dirPath': 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 set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
loadPassages Ran
ran on a synthesized input
microsoft/MSMARCO-Passage-Ranking/generateBERTPassageEmbedding.py
code served (permissive licence) · get_code("4aae25f3615c1cb4")
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 passage data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write("1\tHello World! This is a test passage.\n")
temp_file.write("2\tAnother PASSAGE with SPECIAL@#$%^&*()characters and   multiple spaces.\n")
temp_file.write("3\tthird passage - lowercase and punctuation!!!\n")
temp_file.close()

ARGS = {
    'filename': 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
loadQueries Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateArtificialSessions.py
code served (permissive licence) · get_code("31374e7d57fb5faa")
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 set. 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 sample query data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_filename = temp_file.name

# Write some sample queries to the file
sample_queries = [
    "query one",
    "query two",
    "query three",
    "another query"
]

for query in sample_queries:
    temp_file.write(query + '\n')

temp_file.close()

# Define the ARGS dictionary
ARGS = {
    'filename': temp_filename
}
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 set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
loadQueries Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateQueryEmbeddings.py
code served (permissive licence) · get_code("f65721714fdad34b")
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 tempfile
import os

# Create a temporary file with sample query data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write("query one\n")
temp_file.write("query two\n")
temp_file.write("query three\n")
temp_file.close()

filename = temp_file.name

ARGS = {
    'filename': filename
}
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
loadSessions Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateArtificialSessions.py
code served (permissive licence) · get_code("daf613df77edb1b0")
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. 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 tab-separated session data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_file.write("session1\tuser1\taction1\n")
temp_file.write("session2\tuser2\taction2\taction3\n")
temp_file.write("session3\tuser3\n")
temp_file.close()

ARGS = {
    'filename': 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 list — a list 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
loadVectors Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateArtificialSessions.py
code served (permissive licence) · get_code("695af7f11d756cdd")
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 file with test data
temp_file = tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt')
temp_filename = temp_file.name

# Write test data: query \t space-separated vectors
test_data = [
    "query1\t0.1 0.2 0.3 0.4",
    "query2\t0.5 0.6 0.7 0.8",
    "query3\t0.9 1.0 1.1 1.2",
    "query4\t1.3 1.4 1.5 1.6"
]

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

temp_file.close()

# Define the arguments
realQueries = ["query1", "query3"]
artificialQueries = ["query2", "query4"]

ARGS = {
    "filename": temp_filename,
    "realQueries": realQueries,
    "artificialQueries": artificialQueries
}
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
load_candidate_from_stream Ran
ran; our contract draft was wrong, not the code
microsoft/MSMARCO-Passage-Ranking/ms_marco_eval.py
code served (permissive licence) · get_code("44dabd5275dfdd74")
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 io

# Create a stream with candidate passage data
# Format: query_id\tpassage_id\trank
stream_content = """1\t101\t1
1\t102\t2
1\t103\t3
2\t201\t1
2\t202\t2
3\t301\t1
"""

# Convert string to a file-like object
f = io.StringIO(stream_content)

ARGS = {
    "f": f
}
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_reference_from_stream Ran
ran; our contract draft was wrong, not the code
microsoft/MSMARCO-Passage-Ranking/ms_marco_eval.py
code served (permissive licence) · get_code("b09cad901d138056")
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 io

# The function reads from a stream (file-like object)
# It expects lines with tab-separated values where:
# - l[0] is query_id (int)
# - l[2] is passage_id (int)

stream_content = """1\tpassage_0\t101
1\tpassage_1\t102
2\tpassage_0\t201
3\tpassage_2\t302
3\tpassage_1\t301
"""

f = io.StringIO(stream_content)

ARGS = {
    "f": f
}
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
normalize_answer Ran
ran on a synthesized input
microsoft/OpenKP/evaluate.py
code served (permissive licence) · get_code("ae96a7f4fe58827f")
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 re
import string

def normalize_answer(s):

    def remove_articles(text):
        return re.sub('\\b(a|an|the)\\b', ' ', text)

    def white_space_fix(text):
        return ' '.join(text.split())

    def remove_punc(text):
        exclude = set(string.punctuation)
        return ''.join((ch for ch in text if ch not in exclude))

    def lower(text):
        return text.lower()
    return ' '.join([lower(x) for x in s]).rstrip()


# The function takes a single parameter 's' which is iterable (used in a for loop)
# It should be a sequence of strings that will be lowercased, joined, and processed
s = "The quick brown fox jumps over the lazy dog! What a beautiful day, isn't it?"

ARGS = {
    "s": s
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
precook Ran
ran; our contract draft was wrong, not the code
microsoft/MSMARCO-Question-Answering/Evaluation/bleu_scorer.py
code served (permissive licence) · get_code("ec4448373d72d636")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) 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.

ARGS = {
    "s": "the quick brown fox jumps over the lazy dog",
    "n": 4,
    "out": False
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
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
process Ran
ran on a synthesized input
microsoft/MSMARCO-Passage-Ranking/generateBERTPassageEmbedding.py
code served (permissive licence) · get_code("ab7871c7af226530")
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 numpy as np

torch.manual_seed(0)

# Analyze the function:
# - passageIDs: iterated over with len(), indexed with [i]
# - response: iterated over with len(), indexed with [i][j]
# Both need to support nested indexing and len()

passageIDs = ['passage_1', 'passage_2']
response = [
    [0.95, 0.87, 0.72],
    [0.54, 0.61, 0.89]
]

ARGS = {
    'passageIDs': passageIDs,
    'response': response
}
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
process Ran
ran on a synthesized input
dfcf93/MSMARCO/ConversationalSearch/generateQueryEmbeddingsBERT.py
code served (permissive licence) · get_code("90cfc8bb296d7674")
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

torch.manual_seed(0)

# queryPack should be a list of items (strings or numbers that can be formatted)
# response should be a list of lists, where each inner list contains items that can be formatted
queryPack = ['query1', 'query2']
response = [['result1', 'result2'], ['result3', 'result4', 'result5']]

ARGS = {
    'queryPack': queryPack,
    'response': response
}
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
remove_empty Ran
ran on a synthesized input
microsoft/MSMARCO-Passage-Ranking/getstats.py
code served (permissive licence) · get_code("6f73ec516845e330")
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. 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

torch.manual_seed(0)

# The function remove_empty takes a_list and filters out empty items
# It checks len(item) > 0 for each item, so items need to be length-checkable
# Create a list with mixed empty and non-empty items

ARGS = {
    "a_list": [
        [],                          # empty list
        [1, 2, 3],                  # non-empty list
        torch.tensor([1, 2, 3, 4]), # non-empty tensor
        [],                          # empty list
        np.array([5, 6]),           # non-empty numpy array
        "",                          # empty string
        "hello",                    # non-empty string
    ]
}
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 list — a list 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
remove_empty Ran
ran on a synthesized input
microsoft/OpenKP/evaluate.py
code served (permissive licence) · get_code("113dbc7ab532d6a3")
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
import string

def normalize_answer(s):
    def remove_articles(text):
        return re.sub('\\b(a|an|the)\\b', ' ', text)

    def white_space_fix(text):
        return ' '.join(text.split())

    def remove_punc(text):
        exclude = set(string.punctuation)
        return ''.join((ch for ch in text if ch not in exclude))

    def lower(text):
        return text.lower()
    return ' '.join([lower(x) for x in s]).rstrip()

def remove_empty(a_list):
    new_list = []
    for i in a_list:
        if len(i) > 0:
            if len(i[0]) > 0:
                new_list.append(normalize_answer(i))
    return new_list


# Analyzing the function:
# - remove_empty takes a_list as parameter
# - It iterates through items in a_list
# - For each item i, it checks len(i) > 0
# - Then it checks len(i[0]) > 0, so i must be indexable and i[0] exists
# - If both conditions pass, it calls normalize_answer(i)
# - normalize_answer expects a string-like object that can be iterated
# - It applies string operations (lower, replace, join, etc.)

# So a_list should be a list of items where:
# - Each item has len > 0
# - Each item has a first element with len > 0
# - Each item should ultimately be a string or sequence of strings

# Construct test data: list of non-empty strings/sequences
a_list_value = [
    "The quick brown fox",
    "An example string with articles",
    "A simple test.",
]

# Edge cases: include some empty strings to test filtering
a_list_with_mixed = [
    "Hello world",
    "",  # This will be filtered out (len == 0)
    "The test string",
]

# Build the ARGS dict
ARGS = {
    "a_list": a_list_with_mixed
}
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
remove_punctuation Ran
ran on a synthesized input
microsoft/MSMARCO-Question-Answering/Evaluation/eval_exp.py
code served (permissive licence) · get_code("0a2e4a7e57b0e25a")
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 sys
import unicodedata

tbl = dict.fromkeys((i for i in range(sys.maxunicode) if unicodedata.category(chr(i)).startswith('P')))

def remove_punctuation(text):
    return text.translate(tbl)

# Create a sample text string with punctuation
sample_text = "Hello, World! This is a test. It has punctuation: @#$%."

ARGS = {
    "text": sample_text
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
whether_ynq Ran
ran on a synthesized input
microsoft/MSMARCO-Question-Answering/Utils/exploredata.py
code served (permissive licence) · get_code("5b99609663c1da5c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

def whether_ynq(query):
    YSNO_MAP = set(['is', 'can', 'does', 'are', 'do', 'should', 'did', 'will', 'was', 'has', 'could', 'were', 'would'])
    tokens = query.split(' ')
    head = tokens[0]
    if head in YSNO_MAP and 'or' not in set(tokens):
        return True
    else:
        return False

ARGS = {
    'query': 'is this a question'
}
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 bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_path Not yet run
unverified (harvested, no recorded run)
AmenRa/rank_eval/ranx/io.py
code served (permissive licence) · get_code("cdbcc024a7240d33")
How it ran

no invocation recorded.

get_tmp_path Not yet run
unverified (harvested, no recorded run)
AmenRa/rank_eval/ranx/io.py
code served (permissive licence) · get_code("036a6050bf6f5374")
How it ran

no invocation recorded.

loadJudgements Not yet run
unverified (harvested, no recorded run)
microsoft/OpenKP/agreement.py
code served (permissive licence) · get_code("2e000671b39ad111")
How it ran

no invocation recorded.

loadQuora Not yet run
unverified (harvested, no recorded run)
dfcf93/MSMARCO/ConversationalSearch/generateQuerysets.py
code served (permissive licence) · get_code("16d3a24b2a035ad7")
How it ran

no invocation recorded.

load_reference Not yet run
unverified (harvested, no recorded run)
microsoft/MSMARCO-Passage-Ranking/ms_marco_eval.py
code served (permissive licence) · get_code("52e247dd5467b4bb")
How it ran

no invocation recorded.

Also present in other papers' code

Of 33 samples lifted from this paper's linked repositories, 33 carry a function identity and 0 do not. Of the 33 distinct functions with an identity, 6 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 2 the same month, 6 later) and 27 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
precook 22 other papers
2 dated earlier, 1 the same month, 19 later
1411.4555 (2014-11)
Show and Tell: A Neural Image Caption Generator
1502.03044 (2015-02)
Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
1611.09571 (2016-11)
Predicting Human Eye Fixations via an LSTM-based Saliency Attentive Model
and 19 more not named here
cook_refs 21 other papers
1 dated earlier, 1 the same month, 19 later
1502.03044 (2015-02)
Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
1611.09571 (2016-11)
Predicting Human Eye Fixations via an LSTM-based Saliency Attentive Model
1806.00807 (2018-06)
Learning Semantic Sentence Embeddings using Sequential Pair-wise Discriminator
and 18 more not named here
cook_test 6 other papers
6 later
1901.11504 (2019-01)
Multi-Task Deep Neural Networks for Natural Language Understanding
2109.08365 (2021-09)
CodeQA: A Question Answering Dataset for Source Code Comprehension
2305.18500 (2023-05)
VAST: A Vision-Audio-Subtitle-Text Omni-Modality Foundation Model and Dataset
and 3 more not named here
load_reference 5 other papers
5 later
2110.05789 (2021-10)
Learning Discrete Representations via Constrained Clustering for Effective and Efficient Dense Retrieval
2022.emnlp-main.727 (2022 (year only))
CODER: An efficient framework for improving retrieval through COntextual Document Embedding Reranking
2209.00179 (2022-09)
Universal Vision-Language Dense Retrieval: Learning A Unified Representation Space for Multi-Modal Retrieval
and 2 more not named here
load_candidate_from_stream 4 other papers
4 later
2022.emnlp-main.727 (2022 (year only))
CODER: An efficient framework for improving retrieval through COntextual Document Embedding Reranking
2209.00179 (2022-09)
Universal Vision-Language Dense Retrieval: Learning A Unified Representation Space for Multi-Modal Retrieval
2310.14037 (2023-10)
MARVEL: Unlocking the Multi-Modal Capability of Dense Retrieval via Visual Module Plugin
and 1 more not named here
load_reference_from_stream 4 other papers
4 later
2022.emnlp-main.727 (2022 (year only))
CODER: An efficient framework for improving retrieval through COntextual Document Embedding Reranking
2209.00179 (2022-09)
Universal Vision-Language Dense Retrieval: Learning A Unified Representation Space for Multi-Modal Retrieval
2310.14037 (2023-10)
MARVEL: Unlocking the Multi-Modal Capability of Dense Retrieval via Visual Module Plugin
and 1 more not named here

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 introduce a large scale MAchine Reading COmprehension dataset, which we name MS MARCO. The dataset comprises of 1,010,916 anonymized questions---sampled from Bing's search query logs---each with a human generated answer and 182,669 completely human rewritten generated answers. In addition, the dataset contains 8,841,823 passages---extracted from 3,563,535 web documents retrieved by Bing---that provide the information necessary for curating the natural language answers. A question in the MS MARCO dataset may have multiple answers or no answers at all. Using this dataset, we propose three different tasks with varying levels of difficulty: (i) predict if a question is answerable given a set of context passages, and extract and synthesize the answer as a human would (ii) generate a well-formed answer (if possible) based on the context passages that can be understood with the question and passage context, and finally (iii) rank a set of retrieved passages given a question. The size of the dataset and the fact that the questions are derived from real user search queries distinguishes MS MARCO from other well-known publicly available datasets for machine reading comprehension and question-answering. We believe that the scale and the real-world nature of this dataset makes it attractive for benchmarking machine reading comprehension and question-answering models.

For agents

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

get_harvested_code_for_paper("1611.09268")
get_code_for_paper("1611.09268")
have("1611.09268")

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

curl https://syntology.ai/api/ran/1611.09268.json

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

[![Syntology run record](https://syntology.ai/api/ran/1611.09268.svg)](https://syntology.ai/paper/1611.09268)

Connect an agent — have() is free.