SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2309.04662 · 2023

MADLAD-400: A Multilingual And Document-Level Large Audited Dataset

arXiv · PDF · Open in the Atlas

Code that ran

7 of 7 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 7 have no contract verdict; 0 failures of our instrument; 0 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/2309.04662.json · badge: /api/ran/2309.04662.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
bramiozo/PubScience community (classifier label)
classifier basis not recorded on the link
7 ran · 0 no recorded run
FunctionStatusWhere it lives
SeinenCorpusPrep Ran
ran on a synthesized input
bramiozo/PubScience/src/pubscience/parse/pubtator.py
code served (permissive licence) · get_code("7af1598325b4e1cc")
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
from typing import List, Dict

ARGS = {
    'text': 'This is a medical text with important entities mentioned throughout the document.',
    'title': 'Medical Report',
    'entities': [
        {
            'start': 10,
            'end': 18,
            'text': 'medical',
            'cui': 'C0376313'
        },
        {
            'start': 33,
            'end': 41,
            'text': 'entities',
            'tui': 'T047'
        },
        {
            'start': 60,
            'end': 70,
            'text': 'mentioned',
            'cui': 'C0030971'
        }
    ],
    'sep': '.'
}
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
add_semantic_types Ran
ran on a synthesized input
bramiozo/PubScience/src/pubscience/parse/pubtator.py
code served (permissive licence) · get_code("04acf69863589dc8")
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
from typing import List, Dict

def add_semantic_types(tui: str, tui_sem_map: Dict[str, str]) -> str:
    return tui_sem_map.get(tui, 'Unknown')

# Setup code
torch.manual_seed(0)

# Create a semantic type mapping dictionary
tui_sem_map = {
    'T001': 'Organism',
    'T002': 'Biologic Function',
    'T003': 'Body Part',
    'T004': 'Chemical',
    'T005': 'Disease',
}

# Create arguments
ARGS = {
    'tui': 'T001',
    'tui_sem_map': tui_sem_map,
}
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
compute_entropy Ran
ran on a synthesized input
bramiozo/PubScience/src/pubscience/clean/cleaner.py
code served (permissive licence) · get_code("849de0c8feb5a9d5")
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 math
from collections import Counter

# Create a frequency counter for the compute_entropy function
counter = Counter({
    'a': 10,
    'b': 5,
    'c': 3,
    'd': 2
})

ARGS = {
    'counter': counter
}
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 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
ensure_directory Ran
ran on a synthesized input
bramiozo/PubScience/src/pubscience/retrieve/retrieve_atc.py
code served (permissive licence) · get_code("a1b35978ea402c8c")
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
import tempfile

# The function ensure_directory takes a dir_path parameter
# It checks if the directory exists, creates it if it doesn't, and returns the path
# We need to provide a valid directory path string

# Create a temporary directory path that doesn't exist yet
temp_base = tempfile.gettempdir()
test_dir_path = os.path.join(temp_base, "test_ensure_directory_" + str(os.getpid()))

# Ensure it doesn't exist before passing to the function
if os.path.exists(test_dir_path):
    import shutil
    shutil.rmtree(test_dir_path)

ARGS = {
    "dir_path": test_dir_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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_char_entr Ran
ran on a synthesized input
bramiozo/PubScience/src/pubscience/clean/cleaner.py
code served (permissive licence) · get_code("8b12b005cb939e5b")
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 math
from collections import Counter

def compute_entropy(counter):
    """
    Given a frequency counter of items in a window, compute Shannon entropy.
    """
    total = sum(counter.values())
    if total == 0:
        return 0.0
    entropy = 0.0
    for count in counter.values():
        p = count / total
        entropy -= p * math.log2(p)
    return entropy

def get_char_entr(char_id_seq, window=5, stride=1):
    """
    Compute local entropies over a sliding window, and record the spans (start, end).
    Returns:
        char_id_entr (list[float]): A list of entropy values for each window.
        spans (list[tuple]): Corresponding (start, end) indices for each window.
    """
    char_id_entr = []
    spans = []
    n = len(char_id_seq)
    for start in range(0, n - window + 1, stride):
        end = start + window
        window_ids = char_id_seq[start:end]
        freq_counter = Counter(window_ids)
        e = compute_entropy(freq_counter)
        char_id_entr.append(e)
        spans.append((start, end))
    return (char_id_entr, spans)

# Setup code to construct valid arguments
char_id_seq = [0, 1, 2, 1, 0, 2, 1, 1, 0, 1, 2, 2, 1, 0, 1]
window = 5
stride = 1

ARGS = {
    "char_id_seq": char_id_seq,
    "window": window,
    "stride": stride,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_char_seq Ran
ran on a synthesized input
bramiozo/PubScience/src/pubscience/clean/cleaner.py
code served (permissive licence) · get_code("698b8f9d59280f86")
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 int; 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
import numpy as np

torch.manual_seed(0)

# The function get_char_seq takes a single parameter 'txt' which is a string
# It converts each character to its ASCII/Unicode code using ord()
# So we need to provide a valid text string

txt = "Hello, World!"

ARGS = {
    "txt": txt
}
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, ndim 1, dtype family int — 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
is_valid_atc_code Ran
ran on a synthesized input
bramiozo/PubScience/src/pubscience/retrieve/retrieve_atc.py
code served (permissive licence) · get_code("9154f133c506c97c")
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 re

atc_code_pattern = re.compile('(^[A-Z]$)|(^[A-Z][^a-z][0-9]{1,2}$)|(^[A-Z][^a-z][0-9]{1,2}[A-Z]{1,2}$)|(^[A-Z][^a-z][0-9]{1,2}[A-Z]{1,2}[0-9]{1,2}$)')

def is_valid_atc_code(code):
    return bool(atc_code_pattern.match(code))

# Valid ATC codes to test:
# - Single uppercase letter: "A"
# - Letter + non-letter + 1-2 digits: "A0" (but needs non-letter between), "A01"
# - Letter + non-letter + 1-2 digits + 1-2 uppercase letters: "A01AB"
# - Letter + non-letter + 1-2 digits + 1-2 uppercase letters + 1-2 digits: "A01AB02"

ARGS = {
    "code": "A01AB02"
}
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 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

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 7 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

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 MADLAD-400, a manually audited, general domain 3T token monolingual dataset based on CommonCrawl, spanning 419 languages. We discuss the limitations revealed by self-auditing MADLAD-400, and the role data auditing had in the dataset creation process. We then train and release a 10.7B-parameter multilingual machine translation model on 250 billion tokens covering over 450 languages using publicly available data, and find that it is competitive with models that are significantly larger, and report the results on different domains. In addition, we train a 8B-parameter language model, and assess the results on few-shot translation. We make the baseline models available to the research community.

For agents

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

get_harvested_code_for_paper("2309.04662")
get_code_for_paper("2309.04662")
have("2309.04662")

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

curl https://syntology.ai/api/ran/2309.04662.json

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

[![Syntology run record](https://syntology.ai/api/ran/2309.04662.svg)](https://syntology.ai/paper/2309.04662)

Connect an agent — have() is free.