SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2404.00264 · 2024

DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation

arXiv · PDF · Open in the Atlas

Code that ran

10 of 12 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 0 have no contract verdict; 10 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2404.00264.json · badge: /api/ran/2404.00264.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
arumaekawa/dilm canonical (classifier label)
classifier basis: census
10 ran · 2 no recorded run
FunctionStatusWhere it lives
average Ran
ran; our contract draft was wrong, not the code
arumaekawa/dilm/src/utils.py
code served (permissive licence) · get_code("78560fa4cc79bb8c")
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 float64, ndim 0, 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

# Test case 1: Simple list of numbers
inputs_numbers = [1.0, 2.0, 3.0, 4.0, 5.0]

# Test case 2: List of lists (will be transposed and averaged element-wise)
inputs_lists = [
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0],
    [7.0, 8.0, 9.0]
]

# Test case 3: List of dicts (will average each key across dicts)
inputs_dicts = [
    {'a': 1.0, 'b': 2.0, 'c': 3.0},
    {'a': 4.0, 'b': 5.0, 'c': 6.0},
    {'a': 7.0, 'b': 8.0, 'c': 9.0}
]

# Create ARGS for the first test case (numbers without std)
ARGS = {
    'inputs': inputs_numbers,
    'std': False
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
batch_to_cuda Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
arumaekawa/dilm/src/utils.py
code served (permissive licence) · get_code("657f8371b248b814")
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. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from transformers import BatchEncoding

torch.manual_seed(0)

# Create a small batch with CPU tensors
batch_dict = {
    "input_ids": torch.randint(0, 100, (2, 4)),
    "attention_mask": torch.ones(2, 4, dtype=torch.long),
    "token_type_ids": torch.zeros(2, 4, dtype=torch.long),
}

# Test with plain dict
batch_plain = batch_dict

# Test with BatchEncoding
batch_encoding = BatchEncoding(batch_dict)

# Use the plain dict version for ARGS
ARGS = {
    "batch": batch_plain
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
batch_to_cuda Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
arumaekawa/dilm/src/coreset/coreset_utils.py
code served (permissive licence) · get_code("8ef672b5cb7eb85a")
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. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from transformers import BatchEncoding

torch.manual_seed(0)

# Create a small batch of tensors on CPU
batch_dict = {
    "input_ids": torch.randint(0, 100, (2, 4)),
    "attention_mask": torch.ones((2, 4), dtype=torch.long),
    "token_type_ids": torch.zeros((2, 4), dtype=torch.long),
}

ARGS = {
    "batch": batch_dict
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
configure_optimizer Ran
ran; our contract draft was wrong, not the code
arumaekawa/dilm/src/utils.py
code served (permissive licence) · get_code("235097ddb542453f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn
from torch.optim import SGD, Adam, AdamW

torch.manual_seed(0)

# Create a minimal model that has named_parameters including bias and LayerNorm.weight
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.layer_norm = nn.LayerNorm(3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.layer_norm(x)
        x = self.linear2(x)
        return x

model = MinimalModel()

ARGS = {
    'model': model,
    'lr': 0.001,
    'optimizer_type': 'adamw',
    'scheduler_type': 'linear',
    'weight_decay': 0.01,
    'warmup_ratio': 0.1,
    'num_train_steps': 1000
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
create_train_prompt Ran
ran; our contract draft was wrong, not the code
arumaekawa/dilm/src/evaluator.py
code served (permissive licence) · get_code("8c527dfae28abf79")
How it ran

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

from datasets import Dataset

torch_seed = 0

# Create a minimal Dataset with required fields
data = {
    'sentence1': ['This is great', 'This is bad'],
    'sentence2': ['I love it', 'I hate it'],
    'labels': [0, 1]
}
dataset = Dataset.from_dict(data)

# Define sentence keys that match the dataset
sentence_keys = ['sentence1', 'sentence2']

# Define label dictionary
label_dict = {
    0: 'positive',
    1: 'negative'
}

ARGS = {
    'dataset': dataset,
    'sentence_keys': sentence_keys,
    'label_dict': label_dict
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
datasets — 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
example_to_prompt Ran
ran; our contract draft was wrong, not the code
arumaekawa/dilm/src/evaluator.py
code served (permissive licence) · get_code("5b3995e98e91b41a")
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

# Construct valid arguments for example_to_prompt

# Create a sample example dict with sentence keys and a label
example = {
    'sentence1': '  This is the first sentence.  ',
    'sentence2': '  This is the second sentence.  ',
    'labels': 0
}

# Create a label dictionary mapping label indices to strings
label_dict = {
    0: 'positive',
    1: 'negative',
    2: 'neutral'
}

# Create a list of sentence keys that exist in the example
sentence_keys = ['sentence1', 'sentence2']

# Define ARGS dict with all parameters
ARGS = {
    'example': example,
    'label_dict': label_dict,
    'sentence_keys': sentence_keys,
    'without_label': False
}
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
herding Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
arumaekawa/dilm/src/coreset/herding.py
code served (permissive licence) · get_code("5fd06d143e5ad4be")
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 Dataset. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
from datasets import Dataset
from transformers import PreTrainedModel, PreTrainedTokenizer

torch.manual_seed(0)

# Mock get_embeddings function
def get_embeddings(dataset, model, tokenizer, sentence_keys):
    # Return embeddings of shape (num_samples, embedding_dim)
    return torch.randn(len(dataset), 4)

# Mock l2_dist function
def l2_dist(a, b):
    # a: (1, embedding_dim), b: (1, embedding_dim)
    # Return distances of shape (num_unselected,)
    return torch.norm(a - b, dim=1)

# Patch the functions into the module namespace
import sys
current_module = sys.modules[__name__]
setattr(current_module, 'get_embeddings', get_embeddings)
setattr(current_module, 'l2_dist', l2_dist)

# Create a minimal mock PreTrainedModel
class MockModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.embedding = nn.Linear(10, 4)
    
    def forward(self, input_ids, attention_mask=None):
        return self.embedding(torch.randn(input_ids.shape[0], 10))

# Create a minimal mock PreTrainedTokenizer
class MockTokenizer:
    def __call__(self, texts, padding=True, truncation=True, return_tensors=None):
        batch_size = len(texts) if isinstance(texts, list) else 1
        return {
            'input_ids': torch.randint(0, 100, (batch_size, 10)),
            'attention_mask': torch.ones(batch_size, 10)
        }

# Create a minimal Dataset
dataset_dict = {
    'text': ['sample 1', 'sample 2'],
    'label': [0, 1]
}
dataset = Dataset.from_dict(dataset_dict)

# Create instances
model = MockModel()
tokenizer = MockTokenizer()

ARGS = {
    'dataset': dataset,
    'dpc': 1,
    'model': model,
    'tokenizer': tokenizer,
    'sentence_keys': ['text']
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
datasets, torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Dataset — a Dataset output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Dataset output the recorded ndim 1 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
l2_dist Ran
ran; our contract draft was wrong, not the code
arumaekawa/dilm/src/coreset/coreset_utils.py
code served (permissive licence) · get_code("87e0c5dc751680de")
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 Tensor, 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

torch.manual_seed(0)

# Create source tensor of shape (n, d) where n=2, d=4
src = torch.randn(2, 4)

# Create target tensor of shape (d,) where d=4
tgt = torch.randn(4)

ARGS = {
    "src": src,
    "tgt": tgt,
}
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 Tensor, ndim 1, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:AttributeError)
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
preprocess_for_classification Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
arumaekawa/dilm/src/evaluator.py
code served (permissive licence) · get_code("7dd5cf74e25b707e")
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 torch

torch.manual_seed(0)

# Create small tensors for the function
# logits: shape (batch_size=2, num_classes=4)
logits = torch.randn(2, 4)

# labels: shape (batch_size=2,)
labels = torch.tensor([0, 2])

ARGS = {
    'logits': logits,
    'labels': labels,
}
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 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
random_selection Ran
ran; our contract draft was wrong, not the code
arumaekawa/dilm/src/coreset/random.py
code served (permissive licence) · get_code("6df27b052e7836d0")
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 Dataset. 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
from datasets import Dataset

# Create a simple dataset with some sample data
data = {
    "text": ["sample 1", "sample 2", "sample 3", "sample 4", "sample 5"],
    "label": [0, 1, 0, 1, 0]
}
dataset = Dataset.from_dict(data)

# Define arguments for random_selection function
ARGS = {
    "dataset": dataset,
    "dpc": 3,
    "seed": 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
datasets — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Dataset — a Dataset output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Dataset output the recorded ndim 1 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_embeddings Not yet run
unverified (harvested, no recorded run)
arumaekawa/dilm/src/coreset/coreset_utils.py
code served (permissive licence) · get_code("499bced10bfc04a9")
How it ran

no invocation recorded.

k_centers Not yet run
unverified (harvested, no recorded run)
arumaekawa/dilm/src/coreset/k_centers.py
code served (permissive licence) · get_code("cf0257c37dd1692f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 11 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 0 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
average 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
batch_to_cuda 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
configure_optimizer 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
create_train_prompt 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
example_to_prompt 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
get_embeddings 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
herding 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
k_centers 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
l2_dist 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
preprocess_for_classification 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation
random_selection 1 other paper
1 same year, order not determinable
2024.findings-naacl.199 (2024 (year only))
DiLM: Distilling Dataset into Language Model for Text-level Dataset Distillation

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

Dataset distillation aims to compress a training dataset by creating a small number of informative synthetic samples such that neural networks trained on them perform as well as those trained on the original training dataset. Current text dataset distillation methods create each synthetic sample as a sequence of word embeddings instead of a text to apply gradient-based optimization; however, such embedding-level distilled datasets cannot be used for training other models whose word embedding weights are different from the model used for distillation. To address this issue, we propose a novel text dataset distillation approach, called Distilling dataset into Language Model (DiLM), which trains a language model to generate informative synthetic training samples as text data, instead of directly optimizing synthetic samples. We evaluated DiLM on various text classification datasets and showed that distilled synthetic datasets from DiLM outperform those from current coreset selection methods. DiLM achieved remarkable generalization performance in training different types of models and in-context learning of large language models. Our code will be available at https://github.com/arumaekawa/DiLM.

For agents

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

get_harvested_code_for_paper("2404.00264")
get_code_for_paper("2404.00264")
have("2404.00264")

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

curl https://syntology.ai/api/ran/2404.00264.json

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

[![Syntology run record](https://syntology.ai/api/ran/2404.00264.svg)](https://syntology.ai/paper/2404.00264)

Connect an agent — have() is free.