SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1910.13461 · ACL · 2019

BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension

Marjan Ghazvininejad, Naman Goyal, Mike Lewis, Luke Zettlemoyer, Abdelrahman Mohamed, Ves Stoyanov, Omer Levy, Yinhan Liu

arXiv · PDF · Open in the Atlas

Code that ran

40 of 53 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (7 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 2 violated it and 24 have no contract verdict; 13 failures of our instrument; 13 no recorded run), measured 2026-11-21; 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/1910.13461.json · badge: /api/ran/1910.13461.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
i2r-simmc/i2r-simmc-2020 community (classifier label)
classifier basis not recorded on the link
15 ran · 0 no recorded run
bayer-science-for-a-better-life/data2text-bioleaflets community (classifier label)
classifier basis not recorded on the link
11 ran · 3 no recorded run
qywu/memformers community
Papers with Code archive
5 ran · 5 no recorded run
jiacheng-xu/text-sum-uncertainty community
Papers with Code archive
3 ran · 1 no recorded run
facebookresearch/GENRE community
Papers with Code archive
2 ran · 3 no recorded run
hihellohowareyou/RESREF_Chatbot_data_for_Korean community
Papers with Code archive
1 ran · 1 no recorded run
john-bradshaw/rxn-lm community
Papers with Code archive
1 ran · 0 no recorded run
cosmoquester/2021-dialogue-summary-competition community
Papers with Code archive
1 ran · 0 no recorded run
cosmoquester/transformers-bart-pretrain community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
BartParams Ran
ran on a synthesized input · constructed an object; no result was computed
john-bradshaw/rxn-lm/rxn_lm/train_utils/training_core.py
code served (permissive licence) · get_code("425f7a153da6ce1c")
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 an instance of its own class, BartParams: the run constructed the object and did not call it, so no result was computed. 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
from dataclasses import dataclass, asdict

torch.manual_seed(0)

ARGS = {
    "encoder_layers": 12,
    "encoder_ffn_dim": 4096,
    "encoder_attention_heads": 16,
    "decoder_layers": 12,
    "decoder_ffn_dim": 4096,
    "decoder_attention_heads": 16,
    "encoder_layerdrop": 0.0,
    "decoder_layerdrop": 0.0,
    "activation_function": "gelu",
    "d_model": 1024,
    "dropout": 0.1,
    "attention_dropout": 0.0,
    "activation_dropout": 0.0,
    "init_std": 0.02,
    "classifier_dropout": 0.0,
    "scale_embedding": False,
    "use_cache": True,
    "num_labels": 3,
    "pad_token_id": 1,
    "bos_token_id": 0,
    "eos_token_id": 2,
    "is_encoder_decoder": True,
    "num_return_sequences": 1,
}
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 BartParams — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a BartParams 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
DataCollector Ran
ran on a synthesized input · constructed an object; no result was computed
jiacheng-xu/text-sum-uncertainty/run_model_pegasus.py
code served (permissive licence) · get_code("e9a610ec98650659")
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 an instance of its own class, DataCollector: the run constructed the object and did not call it, so no result was computed. 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
import tempfile

torch.manual_seed(0)

# Create a temporary directory for output
temp_dir = tempfile.mkdtemp()

# Construct ARGS dictionary with all parameters for DataCollector.__init__
ARGS = {
    'full_data': True,
    'cur_dir': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataCollector — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a DataCollector 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
DefaultModule Ran
ran on a synthesized input · constructed an object; no result was computed
cosmoquester/2021-dialogue-summary-competition/summarizer/method/default.py
code served (permissive licence) · get_code("e01d62d03dcd9ca4")
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 an instance of its own class, DefaultModule: the run constructed the object and did not call it, so no result was computed. 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 torch
import torch.nn as nn
from transformers import BartForConditionalGeneration, BartConfig

torch.manual_seed(0)

# Create a minimal BART model
config = BartConfig(
    vocab_size=1000,
    max_position_embeddings=128,
    encoder_layers=2,
    encoder_ffn_dim=256,
    encoder_attention_heads=2,
    decoder_layers=2,
    decoder_ffn_dim=256,
    decoder_attention_heads=2,
    d_model=64,
    pad_token_id=0,
    eos_token_id=2,
    bos_token_id=2,
    forced_bos_token_id=2,
)

model = BartForConditionalGeneration(config)

# Define arguments for DefaultModule
ARGS = {
    'model': model,
    'total_steps': 1000,
    'max_learning_rate': 1e-4,
    'min_learning_rate': 1e-5,
    'warmup_rate': 0.1,
    'model_save_dir': '/tmp/model_saves'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pytorch_lightning, torch, torchmetrics, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DefaultModule — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a DefaultModule 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
MemBartEncoderOutput Ran
ran on a synthesized input · constructed an object; no result was computed
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("54be1be200dfeae7")
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 an instance of its own class, MemBartEncoderOutput: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from typing import Optional, Tuple

torch.manual_seed(0)

# Create minimal tensors for MemBartEncoderOutput
batch_size = 2
sequence_length = 3
hidden_size = 4
num_heads = 2

# last_hidden_state: (batch_size, sequence_length, hidden_size)
last_hidden_state = torch.randn(batch_size, sequence_length, hidden_size)

# hidden_states: tuple of tensors, one for embeddings + one per layer
# Creating 3 layers worth
hidden_states = (
    torch.randn(batch_size, sequence_length, hidden_size),
    torch.randn(batch_size, sequence_length, hidden_size),
    torch.randn(batch_size, sequence_length, hidden_size),
)

# attentions: tuple of tensors (batch_size, num_heads, sequence_length, sequence_length)
attentions = (
    torch.randn(batch_size, num_heads, sequence_length, sequence_length),
    torch.randn(batch_size, num_heads, sequence_length, sequence_length),
)

# memory_states: (batch_size, sequence_length, hidden_size)
memory_states = torch.randn(batch_size, sequence_length, hidden_size)

# memory_writer_attentions: (batch_size, num_heads, sequence_length, sequence_length)
memory_writer_attentions = torch.randn(batch_size, num_heads, sequence_length, sequence_length)

# encoder_attention_mask: (batch_size, sequence_length)
encoder_attention_mask = torch.ones(batch_size, sequence_length, dtype=torch.bool)

# previous_memory_states: (batch_size, sequence_length, hidden_size)
previous_memory_states = torch.randn(batch_size, sequence_length, hidden_size)

ARGS = {
    "last_hidden_state": last_hidden_state,
    "hidden_states": hidden_states,
    "attentions": attentions,
    "memory_states": memory_states,
    "memory_writer_attentions": memory_writer_attentions,
    "encoder_attention_mask": encoder_attention_mask,
    "previous_memory_states": previous_memory_states,
}
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 MemBartEncoderOutput — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MemBartEncoderOutput 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
MemBartSeq2SeqLMOutput Ran
ran on a synthesized input · constructed an object; no result was computed
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("7dd81fe3cbe34bdb")
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 an instance of its own class, MemBartSeq2SeqLMOutput: the run constructed the object and did not call it, so no result was computed. 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)

batch_size = 2
seq_length = 4
vocab_size = 8
hidden_size = 6
num_heads = 2
num_layers = 3
embed_size_per_head = 3
encoder_seq_length = 5

ARGS = {
    "loss": torch.tensor([0.5], dtype=torch.float32),
    "logits": torch.randn(batch_size, seq_length, vocab_size),
    "past_key_values": (
        (
            torch.randn(batch_size, num_heads, seq_length, embed_size_per_head),
            torch.randn(batch_size, num_heads, seq_length, embed_size_per_head),
            torch.randn(batch_size, num_heads, encoder_seq_length, embed_size_per_head),
            torch.randn(batch_size, num_heads, encoder_seq_length, embed_size_per_head),
        ),
        (
            torch.randn(batch_size, num_heads, seq_length, embed_size_per_head),
            torch.randn(batch_size, num_heads, seq_length, embed_size_per_head),
            torch.randn(batch_size, num_heads, encoder_seq_length, embed_size_per_head),
            torch.randn(batch_size, num_heads, encoder_seq_length, embed_size_per_head),
        ),
    ),
    "decoder_hidden_states": (
        torch.randn(batch_size, seq_length, hidden_size),
        torch.randn(batch_size, seq_length, hidden_size),
        torch.randn(batch_size, seq_length, hidden_size),
    ),
    "decoder_attentions": (
        torch.randn(batch_size, num_heads, seq_length, seq_length),
        torch.randn(batch_size, num_heads, seq_length, seq_length),
    ),
    "cross_attentions": (
        torch.randn(batch_size, num_heads, seq_length, encoder_seq_length),
        torch.randn(batch_size, num_heads, seq_length, encoder_seq_length),
    ),
    "encoder_last_hidden_state": torch.randn(batch_size, encoder_seq_length, hidden_size),
    "encoder_hidden_states": (
        torch.randn(batch_size, encoder_seq_length, hidden_size),
        torch.randn(batch_size, encoder_seq_length, hidden_size),
    ),
    "encoder_attentions": (
        torch.randn(batch_size, num_heads, encoder_seq_length, encoder_seq_length),
    ),
    "memory_states": torch.randn(batch_size, seq_length, hidden_size),
}
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 MemBartSeq2SeqLMOutput — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MemBartSeq2SeqLMOutput 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
MemBartSeq2SeqModelOutput Ran
ran on a synthesized input · constructed an object; no result was computed
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("ff3f2f670826701f")
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 an instance of its own class, MemBartSeq2SeqModelOutput: the run constructed the object and did not call it, so no result was computed. 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)

# Define tensor dimensions: batch=2, hidden=4, seq_len=3, num_heads=2, embed_per_head=2
batch_size = 2
seq_len = 3
hidden_size = 4
num_heads = 2
embed_per_head = 2
num_layers = 2

# Create last_hidden_state
last_hidden_state = torch.randn(batch_size, seq_len, hidden_size)

# Create past_key_values: tuple of tuples with self-attention and cross-attention tensors
past_key_values = tuple([
    (
        torch.randn(batch_size, num_heads, seq_len, embed_per_head),  # self-attn key
        torch.randn(batch_size, num_heads, seq_len, embed_per_head),  # self-attn value
        torch.randn(batch_size, num_heads, seq_len, embed_per_head),  # cross-attn key
        torch.randn(batch_size, num_heads, seq_len, embed_per_head),  # cross-attn value
    )
    for _ in range(num_layers)
])

# Create decoder_hidden_states: tuple of tensors (one per layer + embeddings)
decoder_hidden_states = tuple([
    torch.randn(batch_size, seq_len, hidden_size)
    for _ in range(num_layers + 1)
])

# Create decoder_attentions: tuple of attention tensors
decoder_attentions = tuple([
    torch.randn(batch_size, num_heads, seq_len, seq_len)
    for _ in range(num_layers)
])

# Create cross_attentions: tuple of cross-attention tensors
cross_attentions = tuple([
    torch.randn(batch_size, num_heads, seq_len, seq_len)
    for _ in range(num_layers)
])

# Create encoder_last_hidden_state
encoder_last_hidden_state = torch.randn(batch_size, seq_len, hidden_size)

# Create encoder_hidden_states
encoder_hidden_states = tuple([
    torch.randn(batch_size, seq_len, hidden_size)
    for _ in range(num_layers + 1)
])

# Create encoder_attentions
encoder_attentions = tuple([
    torch.randn(batch_size, num_heads, seq_len, seq_len)
    for _ in range(num_layers)
])

# Create memory_states
memory_states = torch.randn(batch_size, seq_len, hidden_size)

ARGS = {
    "last_hidden_state": last_hidden_state,
    "past_key_values": past_key_values,
    "decoder_hidden_states": decoder_hidden_states,
    "decoder_attentions": decoder_attentions,
    "cross_attentions": cross_attentions,
    "encoder_last_hidden_state": encoder_last_hidden_state,
    "encoder_hidden_states": encoder_hidden_states,
    "encoder_attentions": encoder_attentions,
    "memory_states": memory_states,
}
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 MemBartSeq2SeqModelOutput — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a MemBartSeq2SeqModelOutput 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
UpdateGate Ran
ran on a synthesized input · constructed an object; no result was computed
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("a38901f69351322f")
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 an instance of its own class, UpdateGate: the run constructed the object and did not call it, so no result was computed. 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)

ARGS = {
    "input_dim": 8,
    "inner_dim": 16,
    "pooler_dropout": 0.1,
}
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 UpdateGate — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a UpdateGate 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
_expand_mask Ran
ran; our contract draft was wrong, not the code
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("ff8c2ca593924bd0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 4, dtype family float, shape (2, 1, 3, 4) 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)

# Create a small attention mask tensor of shape [bsz, seq_len]
# bsz=2, seq_len=4
mask = torch.randint(0, 2, (2, 4), dtype=torch.float32)

# dtype for expansion
dtype = torch.float32

# Optional tgt_len parameter (can be None or an integer)
tgt_len = 3

ARGS = {
    "mask": mask,
    "dtype": dtype,
    "tgt_len": tgt_len,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (2, 1, 3, 4) — shape (2, 1, 3, 4) 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
add_to_special_tokens Ran
ran; our contract draft was wrong, not the code
hihellohowareyou/RESREF_Chatbot_data_for_Korean/generator/dataset.py
pointer only (licence: NONE) · get_code("390e7ee6a89b3db0")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, hihellohowareyou/RESREF_Chatbot_data_for_Korean when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

auto_detach_to_cpu Ran
ran; our contract draft was wrong, not the code
jiacheng-xu/text-sum-uncertainty/run_model_pegasus.py
code served (permissive licence) · get_code("61295902b9488656")
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 ndarray, ndim 2, dtype family float, shape (2, 4) 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
from typing import Union, List

torch.manual_seed(0)

# Test case 1: Single tensor
tensor_input = torch.randn(2, 4)

# Test case 2: List of tensors
list_of_tensors = [torch.randn(2, 4), torch.randn(3, 5)]

# Test case 3: Nested list of tensors
nested_list = [[torch.randn(2, 4), torch.randn(3, 5)], [torch.randn(1, 6)]]

# Define ARGS with different test cases - we'll use the single tensor case
ARGS = {
    "inp": tensor_input,
    "dtype": np.float32
}
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 ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
calc_banned_bad_words_ids Ran
ran; our contract draft was wrong, not the code
i2r-simmc/i2r-simmc-2020/src/generation_utils.py
code served (permissive licence) · get_code("6b81c6ad0d553004")
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)

# Analyzing the function:
# - prev_input_ids: Iterable[int] - gets iterated over (outer loop), each element is compared with banned_token_seq
#   It appears to be an iterable of sequences (since _tokens_match is called with prev_input_ids_slice)
# - bad_words_ids: Iterable[int] - gets iterated over (inner loop), each element is a sequence
#   Each banned_token_seq needs len() and indexing operations

# The function expects:
# - prev_input_ids to be an iterable of sequences (lists of token ids)
# - bad_words_ids to be an iterable of sequences (lists of token ids)

# Simple test case:
# prev_input_ids: batch of token sequences
prev_input_ids = [
    [1, 2, 3],      # sequence 1
    [4, 5, 6, 7]    # sequence 2
]

# bad_words_ids: sequences of bad tokens to ban
# Each bad_words_ids entry is a sequence, and we ban the last token if the previous tokens match
bad_words_ids = [
    [2, 3, 999],    # ban token 999 if preceded by [2, 3]
    [5, 6, 888],    # ban token 888 if preceded by [5, 6]
    [10, 777]       # ban token 777 if preceded by [10]
]

ARGS = {
    "prev_input_ids": prev_input_ids,
    "bad_words_ids": bad_words_ids
}
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
calc_banned_ngram_tokens Ran
ran on a synthesized input
i2r-simmc/i2r-simmc-2020/src/generation_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from shmsw25/AmbigQA, is pointer only (licence: NONE) · get_code("8d77347cfa15b834")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from shmsw25/AmbigQA, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

calc_section_len Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/helpers.py
code served (permissive licence) · get_code("9b6e878d15f549bc")
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 numpy as np

class Section:
    def __init__(self, content):
        self.section_content = content

class PackageLeaflet:
    def __init__(self, section_contents):
        self.section1 = Section(section_contents[0])
        self.section2 = Section(section_contents[1])
        self.section3 = Section(section_contents[2])
        self.section4 = Section(section_contents[3])
        self.section5 = Section(section_contents[4])
        self.section6 = Section(section_contents[5])

# Create sample package leaflets with varying content lengths
leaflet1 = PackageLeaflet([
    "This is section 1 content",
    "Section 2 with more text here",
    "Section 3 content",
    None,
    "Section 5 text content here",
    "Section 6"
])

leaflet2 = PackageLeaflet([
    "Another section 1",
    None,
    "Section 3 from leaflet 2",
    "Section 4 content",
    "Section 5 longer text content",
    "Section 6 with additional info"
])

leaflet3 = PackageLeaflet([
    "Third leaflet section 1 text",
    "Third leaflet section 2",
    None,
    "Third leaflet section 4",
    None,
    "Third leaflet section 6"
])

package_leaflets = [leaflet1, leaflet2, leaflet3]

ARGS = {
    "package_leaflets": package_leaflets
}
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 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
check_none_documents Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/preparation/clean_text.py
code served (permissive licence) · get_code("6c663245fb8a9853")
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 int, ndim 0, dtype family int, 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.

class Leaflet:
    def __init__(self, id, section1=None, section2=None, section3=None, section4=None, section5=None, section6=None):
        self.id = id
        self.section1 = section1
        self.section2 = section2
        self.section3 = section3
        self.section4 = section4
        self.section5 = section5
        self.section6 = section6

# Create test leaflets
leaflet1 = Leaflet(id=1, section1=None, section2=None, section3=None, section4=None, section5=None, section6=None)
leaflet2 = Leaflet(id=2, section1="content", section2="content", section3="content", section4="content", section5="content", section6="content")
leaflet3 = Leaflet(id=3, section1=None, section2=None, section3=None, section4=None, section5=None, section6=None)
leaflet4 = Leaflet(id=4, section1="some content", section2=None, section3=None, section4=None, section5=None, section6=None)

package_leaflets_raw = [leaflet1, leaflet2, leaflet3, leaflet4]

ARGS = {
    "package_leaflets_raw": package_leaflets_raw
}
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 int, ndim 0, dtype family int, 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
chunk_it Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
facebookresearch/GENRE/genre/hf_model.py
pointer only (licence: NOASSERTION) · get_code("d43a426d78c2dacb")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, facebookresearch/GENRE (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

convert_action Ran
ran on a synthesized input
i2r-simmc/i2r-simmc-2020/src/preprocessing/convert_simmc.py
code served (permissive licence) · get_code("2f97d78f96ebf790")
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

# Construct a minimal turn dict matching what convert_action expects
turn = {
    'turn_idx': 0,
    'raw_assistant_keystrokes': [
        {'image_id': '1'}
    ],
    'transcript_annotated': "[\
        {'intent': 'DA:ASK:GET.price', 'id': 'slot1', 'slots': [{'id': 'INFO.color'}]},\
        {'intent': 'DA:INFORM:GENERAL', 'id': 'slot2', 'slots': []}\
    ]"
}

# Construct mm_state dict with memory_images and database_images
mm_state = {
    'memory_images': [1, 2, 3],
    'database_images': [4, 5, 6]
}

ARGS = {
    'turn': turn,
    'mm_state': mm_state
}
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
extract_doc_info Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/preparation/parse_json.py
code served (permissive licence) · get_code("e2635b61a3c6c441")
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 logging

# Configure logging to suppress output during testing
logging.basicConfig(level=logging.CRITICAL)

# Define test data that matches the structure expected by extract_doc_info
file_data = {
    'attachment': [
        {
            'id': 'doc_12345',
            'link': 'https://example.com/product.pdf',
            'content': 'This is the product content with detailed information.'
        }
    ],
    'product_name': 'Widget Pro (previously Widget Basic)'
}

ARGS = {
    'file_data': file_data
}
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
extract_info_attributes Ran
ran on a synthesized input
i2r-simmc/i2r-simmc-2020/src/preprocessing/convert_simmc.py
code served (permissive licence) · get_code("d46c1839e7291c3f")
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

# Construct a valid turn dict with the required structure
# The function expects 'transcript_annotated' key that contains an eval-able string
# representing a list of annotation objects with 'intent' and 'slots' fields

annotation_list = [
    {
        'intent': 'DA:ASK:GET.name',
        'slots': [
            {'id': 'INFO.name', 'value': 'John'},
            {'id': 'OTHER.field', 'value': 'test'}
        ]
    },
    {
        'intent': 'DA:INFORM:GET.phone',
        'slots': []
    },
    {
        'intent': 'DA:ASK:CHECK',
        'slots': [
            {'id': 'INFO.address', 'value': 'Main St'}
        ]
    },
    {
        'intent': 'DA:INFORM:GET.info',
        'slots': [
            {'id': 'INFO.email', 'value': 'test@example.com'},
            {'id': 'INFO.zip', 'value': '12345'}
        ]
    },
    {
        'intent': 'OTHER.intent',
        'slots': []
    }
]

# Create the turn dict with transcript_annotated as a string representation
turn = {
    'transcript_annotated': str(annotation_list)
}

ARGS = {
    'turn': turn
}
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
extract_package_leaflet Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/preparation/parse_json.py
code served (permissive licence) · get_code("5d81e0ebf38a1261")
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 logging

# Configure logging to suppress warnings during execution
logging.basicConfig(level=logging.ERROR)

# Create valid arguments for extract_package_leaflet
doc_content = """
Some initial content here with multiple    spaces and newlines.
B. PACKAGE LEAFLET

Read all of this leaflet carefully before you start taking this medicine.
Keep this leaflet. You may need to read it again.
If you have any further questions, ask your doctor or pharmacist.
This medicine has been given to you only.
Do not pass it on to others.

Instructions for use:
Take one tablet daily with water.

This leaflet was last revised in January 2024.
"""

product_id = "PROD-12345"

ARGS = {
    "doc_content": doc_content,
    "product_id": product_id,
}
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_args_for_furniture_click Ran
ran on a synthesized input
i2r-simmc/i2r-simmc-2020/src/preprocessing/convert_simmc.py
code served (permissive licence) · get_code("8b437b06f0647a87")
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 json

NEXT_STATE = 'nextState'
PREVIOUS_STATE = 'previousState'
TEXT_PREFABS_CAROUSEL = 'textPrefabsInCarousel'

# Create a valid stroke JSON string that matches the function's expected structure
stroke_dict = {
    NEXT_STATE: {
        TEXT_PREFABS_CAROUSEL: ["text1", "text2", "text3"]
    },
    PREVIOUS_STATE: {
        TEXT_PREFABS_CAROUSEL: ["text1", "text2"]
    }
}

stroke = json.dumps(stroke_dict)

ARGS = {
    "stroke": stroke
}
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
get_config Ran
ran on a synthesized input
i2r-simmc/i2r-simmc-2020/src/util.py
code served (permissive licence) · get_code("05809fee3ec27f95")
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 yaml
import tempfile
import os

# Create a temporary YAML config file
temp_dir = tempfile.mkdtemp()
config_path = os.path.join(temp_dir, 'config.yml')

config_data = {
    'model': 'resnet50',
    'learning_rate': 0.001,
    'batch_size': 32,
    'epochs': 10,
    'layers': [64, 128, 256],
    'debug': True
}

with open(config_path, 'w') as f:
    yaml.dump(config_data, f)

ARGS = {
    'config_path': config_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
yaml — 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
get_random_string Ran
ran; our contract draft was wrong, not the code
jiacheng-xu/text-sum-uncertainty/run_model_pegasus.py
code served (permissive licence) · get_code("47a59a0b45c916c8")
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 string

# Set seed for reproducibility
random.seed(0)

# The function get_random_string takes a single parameter: length (an integer)
# It generates a random string of lowercase letters of the specified length

ARGS = {
    "length": 10
}
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
hf_bucket_url Ran
ran; our contract draft was wrong, not the code
i2r-simmc/i2r-simmc-2020/src/file_utils.py
code served (permissive licence) · get_code("070b5bdeecd71825")
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.

# Setup code for hf_bucket_url function

ARGS = {
    "model_id": "bert-base-uncased",
    "filename": "pytorch_model.bin",
    "use_cdn": True
}
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
invert_mask Ran
ran, violated the contract we drafted
i2r-simmc/i2r-simmc-2020/src/model.py
pointer only (licence: MIT); the stored copy get_code serves, from yeliu918/KG-BART, is pointer only (licence: NONE) · get_code("874a42377fa0659f")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from yeliu918/KG-BART, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

is_remote_url Ran
ran, violated the contract we drafted
i2r-simmc/i2r-simmc-2020/src/file_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from vis-nlp/ChartQA, is pointer only (licence: GPL-3.0) · get_code("899d1e2736236351")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from vis-nlp/ChartQA, and GPL-3.0, as GitHub detects it on that repository, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

load_model Ran
ran on a synthesized input
i2r-simmc/i2r-simmc-2020/src/util.py
code served (permissive licence) · get_code("52e211fe8bfc51fe")
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 OrderedDict. 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 torch
import torch.nn as nn
import tempfile

torch.manual_seed(0)

# Create a temporary directory and a simple model file
temp_dir = tempfile.mkdtemp()
model_path = os.path.join(temp_dir, 'test_model.pt')

# Create a simple model and save it
simple_model = nn.Linear(4, 3)
torch.save(simple_model.state_dict(), model_path)

ARGS = {
    'path': model_path
}
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 OrderedDict — a OrderedDict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a OrderedDict 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
make_padding_mask Ran
ran; our contract draft was wrong, not the code
i2r-simmc/i2r-simmc-2020/src/model.py
pointer only (licence: MIT); the stored copy get_code serves, from yeliu918/KG-BART, is pointer only (licence: NONE) · get_code("f20c9b1aead27fea")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from yeliu918/KG-BART, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

make_valid_mapping Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/helpers.py
code served (permissive licence) · get_code("718d6d148c2907ad")
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

class Section:
    def __init__(self, section_content=None, entity_recognition=None):
        self.section_content = section_content
        self.entity_recognition = entity_recognition

class PackageLeaflet:
    def __init__(self, section_content_list=None, entity_recognition_list=None):
        self.section1 = Section(section_content_list[0] if section_content_list else None, 
                               entity_recognition_list[0] if entity_recognition_list else None)
        self.section2 = Section(section_content_list[1] if section_content_list and len(section_content_list) > 1 else None,
                               entity_recognition_list[1] if entity_recognition_list and len(entity_recognition_list) > 1 else None)
        self.section3 = Section(section_content_list[2] if section_content_list and len(section_content_list) > 2 else None,
                               entity_recognition_list[2] if entity_recognition_list and len(entity_recognition_list) > 2 else None)
        self.section4 = Section(section_content_list[3] if section_content_list and len(section_content_list) > 3 else None,
                               entity_recognition_list[3] if entity_recognition_list and len(entity_recognition_list) > 3 else None)
        self.section5 = Section(section_content_list[4] if section_content_list and len(section_content_list) > 4 else None,
                               entity_recognition_list[4] if entity_recognition_list and len(entity_recognition_list) > 4 else None)
        self.section6 = Section(section_content_list[5] if section_content_list and len(section_content_list) > 5 else None,
                               entity_recognition_list[5] if entity_recognition_list and len(entity_recognition_list) > 5 else None)

torch.manual_seed(0)

leaflet1 = PackageLeaflet(
    section_content_list=["content1", "content2", "content3", "content4", "content5", "content6"],
    entity_recognition_list=[["entity1"], ["entity2"], None, ["entity4"], [], ["entity6"]]
)

leaflet2 = PackageLeaflet(
    section_content_list=["text1", None, "text3", [], "text5", "text6"],
    entity_recognition_list=[["ent1"], ["ent2"], ["ent3"], ["ent4"], ["ent5"], None]
)

leaflet3 = PackageLeaflet(
    section_content_list=[None, "content", "content", "content", "content", "content"],
    entity_recognition_list=[["e1"], None, ["e3"], ["e4"], ["e5"], ["e6"]]
)

package_leaflets = [leaflet1, leaflet2, leaflet3]

ARGS = {
    "package_leaflets": package_leaflets
}
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
post_process_wikidata Ran
ran; our contract draft was wrong, not the code
facebookresearch/GENRE/genre/hf_model.py
pointer only (licence: NOASSERTION) · get_code("548c3509c23676c4")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, facebookresearch/GENRE (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

process_section6 Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/preparation/clean_text.py
code served (permissive licence) · get_code("e683aae8732f6e92")
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.

class Section:
    def __init__(self, section_content):
        self.section_content = section_content

class Leaflet:
    def __init__(self, section6_content):
        self.section6 = Section(section6_content)

# Create test leaflets with various section6 contents
leaflet1 = Leaflet(
    "Some information about the medicine. Marketing Authorisation Holder is Company X, address details here"
)

leaflet2 = Leaflet(
    "Additional details. Marketing Authorization Holder: Company Y, contact info here"
)

leaflet3 = Leaflet(
    "Product info with both Marketing Authorisation Holder and Marketing Authorization Holder references"
)

leaflet4 = Leaflet(None)

leaflet5 = Leaflet("Just some content without the keywords")

package_leaflets_raw = [leaflet1, leaflet2, leaflet3, leaflet4, leaflet5]

ARGS = {
    "package_leaflets_raw": package_leaflets_raw
}
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
process_string Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/preparation/parse_json.py
code served (permissive licence) · get_code("cc1fd572c2800793")
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)

# The function process_string takes a single parameter 'string' which should be a string
# The function strips whitespace, newlines, tabs, and whitespace again from the string
# We need to provide a string with various whitespace characters to test the function

test_string = "  \n\t  hello world  \t\n  "

ARGS = {
    "string": test_string
}
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
reformat_stanza Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/annotations/named_entity_recognition.py
code served (permissive licence) · get_code("13594c848a4f3848")
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.

# Create a stand-in Span class to mimic stanza.models.common.doc.Span
class Span:
    def __init__(self, text, entity_type, start_char, end_char):
        self.text = text
        self.type = entity_type
        self.start_char = start_char
        self.end_char = end_char

# Create a stand-in Section class
class Section:
    def __init__(self, section_content=None, entity_recognition=None):
        self.section_content = section_content
        self.entity_recognition = entity_recognition

# Create a stand-in Leaflet class
class Leaflet:
    def __init__(self):
        self.section1 = Section(section_content="This is section 1 content with some text.", entity_recognition=[Span("aspirin", "MEDICATION", 10, 16)])
        self.section2 = Section(section_content="Section 2 with more information.", entity_recognition=[Span("ibuprofen", "MEDICATION", 15, 24)])
        self.section3 = Section(section_content="Section 3 details.", entity_recognition=None)
        self.section4 = Section(section_content=None, entity_recognition=None)
        self.section5 = Section(section_content="x", entity_recognition=[])
        self.section6 = Section(section_content="Section 6 with entities.", entity_recognition=[Span("paracetamol", "MEDICATION", 12, 23), Span("headache", "SYMPTOM", 30, 38)])

# Create package_leaflets_Stanza with multiple leaflets
package_leaflets_Stanza = [Leaflet(), Leaflet()]

ARGS = {
    "package_leaflets_Stanza": package_leaflets_Stanza
}
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_URLs Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/preparation/clean_text.py
code served (permissive licence) · get_code("1d119e078734e42d")
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

def remove_URLs(section_content):
    """
    Remove URL from the section content

    :param section_content: content of a section
    :return: content of a section without URLs
    """
    section_content = re.sub('http\\S+', '', section_content)
    section_content = re.sub('www\\S+', '', section_content)
    section_content = re.sub('mailto\\S+', '', section_content)
    section_content = re.sub(' +', ' ', section_content)
    return section_content


ARGS = {
    "section_content": "Check out http://example.com and www.google.com also mailto:test@example.com for more info"
}
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
remove_duplicates Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/bioleaflets/helpers.py
code served (permissive licence) · get_code("d0a738e3442f47f8")
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.

class Section:
    def __init__(self, section_content=None, entity_recognition=None):
        self.section_content = section_content
        self.entity_recognition = entity_recognition

class PackageLeaflet:
    def __init__(self):
        self.section1 = Section()
        self.section2 = Section()
        self.section3 = Section()
        self.section4 = Section()
        self.section5 = Section()
        self.section6 = Section()

# Create test data with some duplicates and some None values
leaflet1 = PackageLeaflet()
leaflet1.section1 = Section(
    section_content="Content A",
    entity_recognition=[{"Text": "entity1"}, {"Text": "entity2"}]
)
leaflet1.section2 = Section(
    section_content="Content B",
    entity_recognition=[{"Text": "entity3"}]
)
leaflet1.section3 = Section(
    section_content="Content A",  # Duplicate content
    entity_recognition=[{"Text": "entity1"}, {"Text": "entity2"}]  # Duplicate NER
)
leaflet1.section4 = Section(None, None)  # Both None
leaflet1.section5 = Section(
    section_content="Content C",
    entity_recognition=[{"Text": "entity4"}]
)
leaflet1.section6 = Section(
    section_content="Content D",
    entity_recognition=None  # NER is None
)

leaflet2 = PackageLeaflet()
leaflet2.section1 = Section(
    section_content="Content E",
    entity_recognition=[{"Text": "entity5"}]
)
leaflet2.section2 = Section(
    section_content="Content F",
    entity_recognition=[{"Text": "entity6"}]
)
leaflet2.section3 = Section(
    section_content="Content C",  # Duplicate with leaflet1.section5
    entity_recognition=[{"Text": "entity4"}]  # Duplicate with leaflet1.section5
)
leaflet2.section4 = Section(
    section_content="Content G",
    entity_recognition=[{"Text": "entity7"}]
)
leaflet2.section5 = Section(None, None)
leaflet2.section6 = Section(
    section_content="Content H",
    entity_recognition=[{"Text": "entity8"}]
)

package_leaflets = [leaflet1, leaflet2]

ARGS = {
    "package_leaflets": package_leaflets
}
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
replace_none_entities Ran
ran on a synthesized input
bayer-science-for-a-better-life/data2text-bioleaflets/annotations/named_entity_recognition.py
code served (permissive licence) · get_code("5ab962d0a8a909a1")
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.

class Section:
    def __init__(self, section_content=None, entity_recognition=None):
        self.section_content = section_content
        self.entity_recognition = entity_recognition

class Leaflet:
    def __init__(self):
        self.section1 = Section(section_content="Content 1", entity_recognition=[{"entity": "drug", "value": "aspirin"}])
        self.section2 = Section(section_content="Content 2", entity_recognition=[])
        self.section3 = Section(section_content=None, entity_recognition=None)
        self.section4 = Section(section_content="", entity_recognition=[{"entity": "dose", "value": "100mg"}])
        self.section5 = Section(section_content="Content 5", entity_recognition=None)
        self.section6 = Section(section_content="Content 6", entity_recognition=[])

package_leaflets = [
    Leaflet(),
    Leaflet()
]

ARGS = {
    "package_leaflets": package_leaflets
}
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
replace_return_docstrings Ran
ran; our contract draft was wrong, not the code
i2r-simmc/i2r-simmc-2020/src/file_utils.py
code served (permissive licence) · get_code("67aa808378a82feb")
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 function. 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 re

class MockOutputType:
    """Mock output type class for testing"""
    __module__ = "transformers.models.mock"
    __name__ = "MockOutput"
    __doc__ = """
    Mock output class.
    
    Args:
        logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, vocab_size)`):
            Prediction scores of the language modeling head.
        hidden_states (`torch.FloatTensor`, *optional*):
            Hidden states of the model.
        attentions (`torch.FloatTensor`, *optional*):
            Attention weights.
    """

class TFMockOutputType:
    """Mock TensorFlow output type class for testing"""
    __module__ = "transformers.models.mock"
    __name__ = "TFMockOutput"
    __doc__ = """
    Mock TensorFlow output class.
    
    Args:
        logits (`tf.Tensor` of shape `(batch_size, sequence_length, vocab_size)`):
            Prediction scores of the language modeling head.
        hidden_states (`tf.Tensor`, *optional*):
            Hidden states of the model.
    """

def mock_function():
    """
    A mock function with a Returns placeholder.
    
    Args:
        input_ids: Input tensor
        attention_mask: Attention mask
    
    Returns:
        Placeholder for return documentation
    """
    pass

ARGS = {
    "output_type": MockOutputType,
    "config_class": "MockConfig"
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
shift_tokens_right Ran
ran; our contract draft was wrong, not the code
i2r-simmc/i2r-simmc-2020/src/model.py
pointer only (licence: MIT); the stored copy get_code serves, from yeliu918/KG-BART, is pointer only (licence: NONE) · get_code("e00ccf7847d12b59")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from yeliu918/KG-BART, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

text_infilling Ran
ran; our contract draft was wrong, not the code
cosmoquester/transformers-bart-pretrain/transformers_bart_pretrain/data.py
code served (permissive licence) · get_code("8eadeed331b841fc")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Function. 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 tensorflow as tf

ARGS = {
    'mask_token_id': 103,
    'masking_rate': 0.3,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Function — a Function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Function 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
top_k_top_p_filtering Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
i2r-simmc/i2r-simmc-2020/src/generation_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from upper9527/gerea, is pointer only (licence: NONE) · get_code("0dc219ab6716a127")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from upper9527/gerea, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

use_cuda Ran
ran, honoured the contract we drafted
i2r-simmc/i2r-simmc-2020/src/util.py
code served (permissive licence) · get_code("4ac94d84c40c32cb")
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 2, dtype family float, shape (2, 4) as a later run recorded it. 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, 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)

# The function takes a single parameter 'var' which should have a .cuda() method
# This is typically a torch tensor or nn.Module
# We'll create a small tensor as the argument

var = torch.randn(2, 4)

ARGS = {
    "var": var
}
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 2, dtype family float, shape (2, 4) — shape (2, 4) 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
GENRE Not yet run
unverified (harvested, no recorded run)
facebookresearch/GENRE/genre/hf_model.py
pointer only (licence: NOASSERTION) · get_code("f05aeb96d74a93ac")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, facebookresearch/GENRE (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

GENREHubInterface Not yet run
unverified (harvested, no recorded run)
facebookresearch/GENRE/genre/hf_model.py
pointer only (licence: NOASSERTION) · get_code("a8821d279d5c47ec")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, facebookresearch/GENRE (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

MemBartEncoder Not yet run
unverified (harvested, no recorded run)
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("81708af0925e3ecb")
How it ran

no invocation recorded.

MemBartEncoderAttention Not yet run
unverified (harvested, no recorded run)
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("a70369f5e1d4f10b")
How it ran

no invocation recorded.

MemBartEncoderLayer Not yet run
unverified (harvested, no recorded run)
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("0f787d1f3f22c2bc")
How it ran

no invocation recorded.

MemBartForConditionalGeneration Not yet run
unverified (harvested, no recorded run)
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("a47f8c473e362e5a")
How it ran

no invocation recorded.

MemBartModel Not yet run
unverified (harvested, no recorded run)
qywu/memformers/memformers/models/membart/modeling_membart.py
code served (permissive licence) · get_code("6af3a69713d1961b")
How it ran

no invocation recorded.

SumGen Not yet run
unverified (harvested, no recorded run)
jiacheng-xu/text-sum-uncertainty/run_model_pegasus.py
code served (permissive licence) · get_code("5a651b6ef42f5d35")
How it ran

no invocation recorded.

_GENREHubInterface Not yet run
unverified (harvested, no recorded run)
facebookresearch/GENRE/genre/hf_model.py
pointer only (licence: NOASSERTION) · get_code("20157e68a194d015")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, facebookresearch/GENRE (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

calc_bleu Not yet run
unverified (harvested, no recorded run)
bayer-science-for-a-better-life/data2text-bioleaflets/evaluations/calc_metrics.py
code served (permissive licence) · get_code("2d3ccf3606eee8d7")
How it ran

no invocation recorded.

calc_rouge Not yet run
unverified (harvested, no recorded run)
bayer-science-for-a-better-life/data2text-bioleaflets/evaluations/calc_metrics.py
code served (permissive licence) · get_code("03c622fbd053b445")
How it ran

no invocation recorded.

calc_sacrebleu Not yet run
unverified (harvested, no recorded run)
bayer-science-for-a-better-life/data2text-bioleaflets/evaluations/calc_metrics.py
code served (permissive licence) · get_code("e3c82cdc2062e229")
How it ran

no invocation recorded.

generatorDataset Not yet run
unverified (harvested, no recorded run)
hihellohowareyou/RESREF_Chatbot_data_for_Korean/generator/dataset.py
pointer only (licence: NONE) · get_code("350f2fc5631039a2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, hihellohowareyou/RESREF_Chatbot_data_for_Korean when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 53 samples lifted from this paper's linked repositories, 53 carry a function identity and 0 do not. Of the 53 distinct functions with an identity, 13 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 13 later) and 40 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
is_remote_url 14 other papers
1 dated earlier, 13 later
1905.00641 (2019-05)
RetinaFace: Single-stage Dense Face Localisation in the Wild
2004.10964 (2020-04)
Don't Stop Pretraining: Adapt Language Models to Domains and Tasks
2104.08066 (2021-04)
Effect of Visual Extensions on Natural Language Understanding in Vision-and-Language Models
and 11 more not named here
shift_tokens_right 11 other papers
11 later
2009.12005 (2020-09)
MinTL: Minimalist Transfer Learning for Task-Oriented Dialogue Systems
2009.12677 (2020-09)
KG-BART: Knowledge Graph-Augmented BART for Generative Commonsense Reasoning
2021.acl-long.390 (2021 (year only))
Few-Shot Text Ranking with Meta Adapted Synthetic Weak Supervision
and 8 more not named here
invert_mask 10 other papers
10 later
2009.12005 (2020-09)
MinTL: Minimalist Transfer Learning for Task-Oriented Dialogue Systems
2009.12677 (2020-09)
KG-BART: Knowledge Graph-Augmented BART for Generative Commonsense Reasoning
2021.acl-long.390 (2021 (year only))
Few-Shot Text Ranking with Meta Adapted Synthetic Weak Supervision
and 7 more not named here
make_padding_mask 9 other papers
9 later
2009.12005 (2020-09)
MinTL: Minimalist Transfer Learning for Task-Oriented Dialogue Systems
2009.12677 (2020-09)
KG-BART: Knowledge Graph-Augmented BART for Generative Commonsense Reasoning
2021.acl-long.390 (2021 (year only))
Few-Shot Text Ranking with Meta Adapted Synthetic Weak Supervision
and 6 more not named here
calc_banned_ngram_tokens 8 other papers
8 later
2106.01317 (2021-06)
Enriching Transformers with Structured Tensor-Product Representations for Abstractive Summarization
2109.04096 (2021-09)
A Three-Stage Learning Framework for Low-Resource Knowledge-Grounded Dialogue Generation
2109.08828 (2021-09)
Perspective-taking and Pragmatics for Generating Empathetic Responses Focused on Emotion Causes
and 5 more not named here
top_k_top_p_filtering 5 other papers
5 later
2109.04096 (2021-09)
A Three-Stage Learning Framework for Low-Resource Knowledge-Grounded Dialogue Generation
2309.01017 (2023-09)
Contrastive Grouping with Transformer for Referring Image Segmentation
2402.02503 (2024-02)
GeReA: Question-Aware Prompt Captions for Knowledge-based Visual Question Answering
and 2 more not named here
replace_return_docstrings 4 other papers
4 later
2104.08066 (2021-04)
Effect of Visual Extensions on Natural Language Understanding in Vision-and-Language Models
2210.00312 (2022-10)
Multimodal Analogical Reasoning over Knowledge Graphs
2305.01938 (2023-05)
Doc2SoarGraph: Discrete Reasoning over Visually-Rich Table-Text Documents via Semantic-Oriented Hierarchical Graphs
and 1 more not named here
_expand_mask
brick of this paper's assembly: helper
3 other papers
3 later
2210.00312 (2022-10)
Multimodal Analogical Reasoning over Knowledge Graphs
2308.07648 (2023-08)
Prompt Switch: Efficient CLIP Adaptation for Text-Video Retrieval
2506.13366 (2025-06)
Enhancing Goal-oriented Proactive Dialogue Systems via Consistency Reflection and Correction
calc_banned_bad_words_ids 2 other papers
2 later
2109.04096 (2021-09)
A Three-Stage Learning Framework for Low-Resource Knowledge-Grounded Dialogue Generation
2410.03968 (2024-10)
Decoding Game: On Minimax Optimality of Heuristic Text Generation Strategies
get_random_string
brick of this paper's assembly: helper
2 other papers
2 later
2302.02607 (2023-02)
Target-based Surrogates for Stochastic Optimization
2304.10532 (2023-04)
Nerfbusters: Removing Ghostly Artifacts from Casually Captured NeRFs
hf_bucket_url 2 other papers
2 later
2109.08133 (2021-09)
Phrase Retrieval Learns Passage Retrieval, Too
2309.01017 (2023-09)
Contrastive Grouping with Transformer for Referring Image Segmentation
use_cuda 2 other papers
2 later
1911.02707 (2019-11)
Grounded Conversation Generation as Guided Traverses in Commonsense Knowledge Graphs
2110.14266 (2021-10)
SQALER: Scaling Question Answering by Decoupling Multi-Hop and Logical Reasoning
chunk_it
brick of this paper's assembly: helper
1 other paper
1 later
2208.07652 (2022-08)
CorpusBrain: Pre-train a Generative Retrieval Model for Knowledge-Intensive Language Tasks

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 2026-11-21.

Repositories linked to this paper

More are linked than shown here.

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We present BART, a denoising autoencoder for pretraining sequence-to-sequence models. BART is trained by ( 1) corrupting text with an arbitrary noising function, and (2) learning a model to reconstruct the original text. It uses a standard Tranformer-based neural machine translation architecture which, despite its simplicity, can be seen as generalizing BERT (due to the bidirectional encoder), GPT (with the left-to-right decoder), and many other more recent pretraining schemes. We evaluate a number of noising approaches, finding the best performance by both randomly shuffling the order of the original sentences and using a novel in-filling scheme, where spans of text are replaced with a single mask token. BART is particularly effective when fine tuned for text generation but also works well for comprehension tasks. It matches the performance of RoBERTa with comparable training resources on GLUE and SQuAD, achieves new stateof-the-art results on a range of abstractive dialogue, question answering, and summarization tasks, with gains of up to 6 ROUGE. BART also provides a 1.1 BLEU increase over a back-translation system for machine translation, with only target language pretraining. We also report ablation experiments that replicate other pretraining schemes within the BART framework, to better measure which factors most influence end-task performance.

For agents

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

get_harvested_code_for_paper("1910.13461")
get_code_for_paper("1910.13461")
have("1910.13461")

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

curl https://syntology.ai/api/ran/1910.13461.json

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

[![Syntology run record](https://syntology.ai/api/ran/1910.13461.svg)](https://syntology.ai/paper/1910.13461)

Connect an agent — have() is free.