SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2205.12487 · 2022

End-to-End Multimodal Fact-Checking and Explanation Generation: A Challenging Dataset and Models

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 0 failures of our instrument; 1 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2205.12487.json · badge: /api/ran/2205.12487.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
VT-NLP/Mocheg canonical (classifier label)
classifier basis not recorded on the link
4 ran · 1 no recorded run
FunctionStatusWhere it lives
freeze_part_bert Ran
ran on a synthesized input
VT-NLP/Mocheg/retrieval/training/train_bi_encoder_mnrl_mocheg.py
code served (permissive licence) · get_code("e56efd0c650472b9")
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 torch.nn as nn

torch.manual_seed(0)

class TextModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.layer1 = nn.Linear(4, 4)
        self.layer2 = nn.Linear(4, 4)
        self.layer3 = nn.Linear(4, 4)
    
    def forward(self, x):
        return x

class VisionModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.layer1 = nn.Linear(4, 4)
        self.layer2 = nn.Linear(4, 4)
        self.layer3 = nn.Linear(4, 4)
    
    def forward(self, x):
        return x

class InnerModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.text_model = TextModel()
        self.vision_model = VisionModel()
    
    def forward(self, x):
        return x

class OuterModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.model = InnerModel()
    
    def forward(self, x):
        return x

model = OuterModel()
model_wrapped = [model]

ARGS = {
    "model": model_wrapped,
    "freeze_text_layer_num": 1,
    "freeze_img_layer_num": 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 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
freeze_part_bert Ran
ran on a synthesized input
VT-NLP/Mocheg/controllable/classification/model.py
code served (permissive licence) · get_code("aeb538304da4ca97")
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 SimpleBertModel. 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)

class SimpleBertModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.embedding = nn.Linear(10, 8)
        self.layer1 = nn.Linear(8, 6)
        self.layer2 = nn.Linear(6, 4)
        self.layer3 = nn.Linear(4, 2)
    
    def forward(self, x):
        x = self.embedding(x)
        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        return x

bert_model = SimpleBertModel()
freeze_layer_num = 2

ARGS = {
    "bert_model": bert_model,
    "freeze_layer_num": freeze_layer_num,
}
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 SimpleBertModel — a SimpleBertModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleBertModel output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_qrels Ran
ran on a synthesized input
VT-NLP/Mocheg/retrieval/eval/eval_msmarco_mocheg.py
code served (permissive licence) · get_code("188ada4a117ab111")
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 pandas as pd
import os
import tempfile

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

# Create a sample CSV file for testing
sample_data = {
    'TOPIC': ['q1', 'q1', 'q2', 'q2', 'q3'],
    'DOCUMENT#': ['d1', 'd2', 'd3', 'd4', 'd5'],
    'RELEVANCY': [1, 0, 1, 0, 1]
}
df = pd.DataFrame(sample_data)
csv_path = os.path.join(temp_dir, 'text_evidence_qrels_sentence_level.csv')
df.to_csv(csv_path, encoding='utf8', index=False)

ARGS = {
    'data_folder': temp_dir,
    'media': 'txt'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas, tqdm — 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
load_queries Ran
ran on a synthesized input
VT-NLP/Mocheg/retrieval/eval/eval_msmarco_mocheg.py
code served (permissive licence) · get_code("d584c95824703a19")
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 pandas as pd
import os
import tempfile
import tqdm

# Create a temporary directory and CSV file for testing
temp_dir = tempfile.mkdtemp()
csv_file = os.path.join(temp_dir, 'Corpus2.csv')

# Create a sample CSV with the expected columns
df = pd.DataFrame({
    'claim_id': ['claim_1', 'claim_2', 'claim_3', 'claim_4'],
    'Claim': ['This is the first claim', 'This is the second claim', 'This is the third claim', 'This is the fourth claim']
})
df.to_csv(csv_file, index=False, encoding='utf8')

# Define the needed_qids set - a subset of claim_ids to filter
needed_qids = {'claim_1', 'claim_3'}

ARGS = {
    'data_folder': temp_dir,
    'needed_qids': needed_qids
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_top_1000_corpus Not yet run
unverified (harvested, no recorded run)
VT-NLP/Mocheg/retrieval/eval/eval_cross_encoder_mocheg.py
code served (permissive licence) · get_code("48e225b8ebde9eac")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 5 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

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

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

We propose end-to-end multimodal fact-checking and explanation generation, where the input is a claim and a large collection of web sources, including articles, images, videos, and tweets, and the goal is to assess the truthfulness of the claim by retrieving relevant evidence and predicting a truthfulness label (e.g., support, refute or not enough information), and to generate a statement to summarize and explain the reasoning and ruling process. To support this research, we construct Mocheg, a large-scale dataset consisting of 15,601 claims where each claim is annotated with a truthfulness label and a ruling statement, and 33,880 textual paragraphs and 12,112 images in total as evidence. To establish baseline performances on Mocheg, we experiment with several state-of-the-art neural architectures on the three pipelined subtasks: multimodal evidence retrieval, claim verification, and explanation generation, and demonstrate that the performance of the state-of-the-art end-to-end multimodal fact-checking does not provide satisfactory outcomes. To the best of our knowledge, we are the first to build the benchmark dataset and solutions for end-to-end multimodal fact-checking and explanation generation. The dataset, source code and model checkpoints are available at https://github.com/VT-NLP/Mocheg.

For agents

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

get_harvested_code_for_paper("2205.12487")
get_code_for_paper("2205.12487")
have("2205.12487")

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

curl https://syntology.ai/api/ran/2205.12487.json

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

[![Syntology run record](https://syntology.ai/api/ran/2205.12487.svg)](https://syntology.ai/paper/2205.12487)

Connect an agent — have() is free.