SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2008.06048 · 2020

MMM : Exploring Conditional Multi-Track Music Generation with the Transformer

arXiv · PDF · Open in the Atlas

Code that ran

9 of 9 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 9 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/2008.06048.json · badge: /api/ran/2008.06048.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
AI-Guru/MMM-JSB community (classifier label)
classifier basis not recorded on the link
9 ran · 0 no recorded run
FunctionStatusWhere it lives
create_logger Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/logging.py
code served (permissive licence) · get_code("8cf31e4865906735")
How it ran

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

import logging

loggers_dict = {}

def create_logger(name: str):
    global loggers_dict
    if name in loggers_dict:
        return loggers_dict[name]
    else:
        logger = logging.getLogger(name)
        loggers_dict[name] = logger
        logger.setLevel(logging.DEBUG)
        handler = logging.StreamHandler()
        handler.setLevel(logging.DEBUG)
        formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
        handler.setFormatter(formatter)
        logger.addHandler(handler)
        logger.propagate = False
        return logger

ARGS = {
    "name": "test_logger"
}
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 Logger — a Logger output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Logger output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
encode_song_data Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/preprocess/encode.py
code served (permissive licence) · get_code("129598d1deb4afec")
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 numpy as np
import random

random.seed(0)
np.random.seed(0)

# Construct track data with proper structure
def create_track_data(track_number, is_drums=False, num_bars=8):
    bars = []
    for bar_idx in range(num_bars):
        events = []
        if bar_idx % 3 != 0:  # Most bars have events
            events = [
                {'type': 'NOTE_ON', 'pitch': 60},
                {'type': 'TIME_DELTA', 'delta': 120},
                {'type': 'NOTE_OFF', 'pitch': 60},
                {'type': 'NOTE_ON', 'pitch': 64},
                {'type': 'TIME_DELTA', 'delta': 120},
                {'type': 'NOTE_OFF', 'pitch': 64},
            ]
        else:  # Some bars are fills
            events = 'bar_fill'
        
        bars.append({'events': events})
    
    return {
        'number': track_number,
        'drums': is_drums,
        'bars': bars
    }

# Create song data with multiple tracks
song_data = {
    'tracks': [
        create_track_data(0, is_drums=False, num_bars=8),
        create_track_data(1, is_drums=True, num_bars=8),
        create_track_data(2, is_drums=False, num_bars=8),
    ]
}

# Set up parameters for encode_song_data
transpositions = [0, 5]
permute = True
window_size_bars = 4
hop_length_bars = 2
density_bins = np.array([2, 4, 6, 8, 10])
bar_fill = True

ARGS = {
    'song_data': song_data,
    'transpositions': transpositions,
    'permute': permute,
    'window_size_bars': window_size_bars,
    'hop_length_bars': hop_length_bars,
    'density_bins': density_bins,
    'bar_fill': bar_fill
}
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 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
encode_songs_data Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/preprocess/encode.py
code served (permissive licence) · get_code("c0aa9297b2b8818c")
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 numpy as np
import random

random.seed(0)
np.random.seed(0)

# Helper function to create minimal song_data structure
def create_song_data(num_tracks=2, num_bars=8, num_events_per_bar=3):
    tracks = []
    for track_idx in range(num_tracks):
        bars = []
        for bar_idx in range(num_bars):
            events = []
            for event_idx in range(num_events_per_bar):
                event_type = random.choice(['NOTE_ON', 'NOTE_OFF', 'TIME_DELTA'])
                if event_type == 'NOTE_ON' or event_type == 'NOTE_OFF':
                    events.append({
                        'type': event_type,
                        'pitch': random.randint(60, 84)
                    })
                else:  # TIME_DELTA
                    events.append({
                        'type': event_type,
                        'delta': random.randint(1, 480)
                    })
            bars.append({
                'events': events
            })
        tracks.append({
            'number': track_idx,
            'drums': track_idx == 0,  # First track is drums
            'bars': bars
        })
    return {'tracks': tracks}

# Create multiple songs
songs_data = [
    create_song_data(num_tracks=2, num_bars=8, num_events_per_bar=3),
    create_song_data(num_tracks=2, num_bars=8, num_events_per_bar=3)
]

# Define arguments for encode_songs_data
ARGS = {
    'songs_data': songs_data,
    'transpositions': [0, 2, -2],  # List of transposition values
    'permute': True,  # Whether to permute tracks
    'window_size_bars': 4,  # Window size in bars
    'hop_length_bars': 2,  # Hop length in bars
    'density_bins': np.array([5, 10, 15, 20]),  # Density bins for np.digitize
    'bar_fill': False  # Whether to include bar fill
}
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 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
encode_track_data Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/preprocess/encode.py
code served (permissive licence) · get_code("33e98f4a06cc60b9")
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 numpy as np

# Define helper function that's used by encode_track_data
def encode_event_data(event_data, transposition):
    if event_data['type'] == 'NOTE_ON':
        return event_data['type'] + '=' + str(event_data['pitch'] + transposition)
    elif event_data['type'] == 'NOTE_OFF':
        return event_data['type'] + '=' + str(event_data['pitch'] + transposition)
    elif event_data['type'] == 'TIME_DELTA':
        return event_data['type'] + '=' + str(event_data['delta'])

def encode_bar_data(bar_data, transposition, bar_fill=False):
    tokens = []
    if not bar_fill:
        tokens += ['BAR_START']
    else:
        tokens += ['FILL_START']
    if bar_data['events'] == 'bar_fill':
        tokens += ['FILL_IN']
    else:
        for event_data in bar_data['events']:
            tokens += [encode_event_data(event_data, transposition)]
    if not bar_fill:
        tokens += ['BAR_END']
    else:
        tokens += ['FILL_END']
    return tokens

# Create test data
track_data = {
    'number': 5,
    'drums': False,
    'bars': [
        {
            'events': [
                {'type': 'NOTE_ON', 'pitch': 60},
                {'type': 'TIME_DELTA', 'delta': 480},
                {'type': 'NOTE_OFF', 'pitch': 60},
            ]
        },
        {
            'events': [
                {'type': 'NOTE_ON', 'pitch': 64},
                {'type': 'TIME_DELTA', 'delta': 240},
                {'type': 'NOTE_OFF', 'pitch': 64},
            ]
        },
        {
            'events': 'bar_fill'
        }
    ]
}

density_bins = np.array([2, 5, 10, 20])
bar_start_index = 0
bar_end_index = 2
transposition = 12

ARGS = {
    'track_data': track_data,
    'density_bins': density_bins,
    'bar_start_index': bar_start_index,
    'bar_end_index': bar_end_index,
    'transposition': transposition
}
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 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
events_to_events_data Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/preprocess/preprocessutilities.py
code served (permissive licence) · get_code("5920e9d4e94038f4")
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
import numpy as np

def events_to_events_data(events):
    events = sorted(events, key=lambda event: event[2])
    events_data = []
    for event_index, event, event_next in zip(range(len(events)), events, events[1:] + [None]):
        if event_index == 0 and event[2] != 0.0:
            event_data = {'type': 'TIME_DELTA', 'delta': event[2]}
            events_data += [event_data]
        event_data = {'type': event[0], 'pitch': event[1]}
        events_data += [event_data]
        if event_next is None:
            continue
        delta = event_next[2] - event[2]
        assert delta >= 0, events
        if delta != 0.0:
            event_data = {'type': 'TIME_DELTA', 'delta': delta}
            events_data += [event_data]
    return events_data

# Construct valid events for the function
# Each event is a tuple: (type_str, pitch_value, time_value)
events = [
    ('NOTE_ON', 60, 0.0),
    ('NOTE_OFF', 60, 0.5),
    ('NOTE_ON', 64, 1.0),
    ('NOTE_OFF', 64, 1.5),
]

ARGS = {
    'events': events
}
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 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
generate Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/helpers/samplinghelpers.py
code served (permissive licence) · get_code("573835f67b3e07a7")
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

torch.manual_seed(0)

# Stand-in tokenizer class with encode and decode methods
class SimpleTokenizer:
    def __init__(self, vocab_size=100):
        self.vocab_size = vocab_size
    
    def encode(self, text, return_tensors=None):
        # Convert text to token ids
        token_ids = [hash(text) % self.vocab_size for _ in range(5)]
        tensor = torch.tensor([token_ids], dtype=torch.long)
        if return_tensors == 'pt':
            return tensor
        return token_ids
    
    def decode(self, token_ids):
        # Convert token ids back to text
        if isinstance(token_ids, torch.Tensor):
            token_ids = token_ids.tolist()
        return "generated text sample"

# Stand-in model class with generate method
class SimpleModel(nn.Module):
    def __init__(self, vocab_size=100, hidden_size=8):
        super().__init__()
        self.vocab_size = vocab_size
        self.hidden_size = hidden_size
        self.embedding = nn.Embedding(vocab_size, hidden_size)
        self.linear = nn.Linear(hidden_size, vocab_size)
    
    def generate(self, input_ids, max_length=1000, temperature=0.9):
        # Simulate generation by returning the input_ids padded with random tokens
        batch_size = input_ids.shape[0]
        # Return a tensor of shape (batch_size, some_length) with token ids
        generated = torch.randint(0, self.vocab_size, (batch_size, 10), dtype=torch.long)
        return generated
    
    def forward(self, input_ids):
        embeddings = self.embedding(input_ids)
        logits = self.linear(embeddings)
        return logits

# Create instances
model = SimpleModel(vocab_size=100, hidden_size=8)
tokenizer = SimpleTokenizer(vocab_size=100)
token_sequence = "hello world"

ARGS = {
    "model": model,
    "tokenizer": tokenizer,
    "token_sequence": token_sequence
}
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_priming_token_sequence Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/helpers/samplinghelpers.py
code served (permissive licence) · get_code("0d3f35802ce765a7")
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 random
import tempfile
import os

# Create a temporary data file with sample token sequences
temp_dir = tempfile.mkdtemp()
data_path = os.path.join(temp_dir, "tokens.txt")

# Write sample data with TRACK_END tokens
with open(data_path, 'w') as f:
    f.write("token1 token2 TRACK_END token3 token4 TRACK_END token5\n")
    f.write("hello world TRACK_END foo bar TRACK_END baz\n")
    f.write("a b c TRACK_END d e f TRACK_END g\n")

ARGS = {
    "data_path": data_path,
    "stop_on_track_end": None,
    "stop_after_n_tokens": None,
    "return_original": False,
}
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
note_sequence_to_bars Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/helpers/noteseqhelpers.py
code served (permissive licence) · get_code("0aaab2ae2d7b40a7")
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
import numpy as np

class Note:
    def __init__(self, start_time, pitch, velocity, end_time):
        self.start_time = start_time
        self.pitch = pitch
        self.velocity = velocity
        self.end_time = end_time

class Tempo:
    def __init__(self, qpm):
        self.qpm = qpm

class NoteSequence:
    def __init__(self, tempos, notes, total_time):
        self.tempos = tempos
        self.notes = notes
        self.total_time = total_time

# Create a small stand-in NoteSequence
tempos = [Tempo(qpm=120.0)]
notes = [
    Note(start_time=0.0, pitch=60, velocity=80, end_time=0.5),
    Note(start_time=0.5, pitch=64, velocity=80, end_time=1.0),
    Note(start_time=2.0, pitch=67, velocity=80, end_time=2.5),
    Note(start_time=2.5, pitch=72, velocity=80, end_time=3.0),
]
total_time = 4.0

note_sequence = NoteSequence(tempos=tempos, notes=notes, total_time=total_time)
threshold = 0.1

ARGS = {
    "note_sequence": note_sequence,
    "threshold": threshold,
}
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 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
set_note_sequence_tempo Ran
ran on a synthesized input
AI-Guru/MMM-JSB/source/helpers/noteseqhelpers.py
code served (permissive licence) · get_code("08b4219bbb9aebcc")
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 NoteSequence. 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.

class Tempo:
    def __init__(self, time=0.0, qpm=120.0):
        self.time = time
        self.qpm = qpm

class Note:
    def __init__(self, start_time=0.0, end_time=1.0, pitch=60, velocity=64):
        self.start_time = start_time
        self.end_time = end_time
        self.pitch = pitch
        self.velocity = velocity

class NoteSequence:
    def __init__(self, tempos=None, notes=None, total_time=4.0):
        self.tempos = tempos if tempos is not None else []
        self.notes = notes if notes is not None else []
        self.total_time = total_time

note_sequence = NoteSequence(
    tempos=[Tempo(time=0.0, qpm=120.0)],
    notes=[
        Note(start_time=0.0, end_time=1.0, pitch=60, velocity=64),
        Note(start_time=1.0, end_time=2.0, pitch=64, velocity=64),
    ],
    total_time=4.0
)

target_tempo = 60.0

ARGS = {
    "note_sequence": note_sequence,
    "target_tempo": target_tempo,
}
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 NoteSequence — a NoteSequence output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a NoteSequence 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

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 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 9 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 propose the Multi-Track Music Machine (MMM), a generative system based on the Transformer architecture that is capable of generating multi-track music. In contrast to previous work, which represents musical material as a single time-ordered sequence, where the musical events corresponding to different tracks are interleaved, we create a time-ordered sequence of musical events for each track and concatenate several tracks into a single sequence. This takes advantage of the Transformer's attention-mechanism, which can adeptly handle long-term dependencies. We explore how various representations can offer the user a high degree of control at generation time, providing an interactive demo that accommodates track-level and bar-level inpainting, and offers control over track instrumentation and note density.

For agents

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

get_harvested_code_for_paper("2008.06048")
get_code_for_paper("2008.06048")
have("2008.06048")

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

curl https://syntology.ai/api/ran/2008.06048.json

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

[![Syntology run record](https://syntology.ai/api/ran/2008.06048.svg)](https://syntology.ai/paper/2008.06048)

Connect an agent — have() is free.