SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.05707 · NeurIPS · 2021

FEVEROUS: Fact Extraction and VERification Over Unstructured and Structured information

Michael Schlichtkrull, Rami Aly, Zhijiang Guo, James Thorne, Andreas Vlachos, Christos Christodoulopoulos, Oana Cocarascu, Arpit Facebook, Roberto Fico, Lewis Ford, Lee Leighton, Martin Berkeley

arXiv · PDF · Open in the Atlas

Code that ran

7 of 8 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 6 have no contract verdict; 1 failure 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/2106.05707.json · badge: /api/ran/2106.05707.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
Raldir/FEVEROUS canonical (classifier label)
classifier basis: census
7 ran · 1 no recorded run
FunctionStatusWhere it lives
hash Ran
ran on a synthesized input
Raldir/FEVEROUS/src/feverous/database/utils.py
code served (permissive licence) · get_code("fe524a16685ba8c3")
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.

from sklearn.utils import murmurhash3_32

def hash(token, num_buckets):
    """Unsigned 32 bit murmurhash for feature hashing."""
    return murmurhash3_32(token, positive=True) % num_buckets

ARGS = {
    "token": "example_token",
    "num_buckets": 10
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
sklearn — 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
is_correct_label Ran
ran on a synthesized input
Raldir/FEVEROUS/src/feverous/evaluation/feverous_scorer.py
code served (permissive licence) · get_code("f68c97519ec5138b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    'instance': {
        'label': 'correct',
        'predicted_label': 'CORRECT'
    }
}
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 bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
is_strictly_correct Ran
ran on a synthesized input
Raldir/FEVEROUS/src/feverous/evaluation/feverous_scorer.py
code served (permissive licence) · get_code("4c72e43932e29578")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

# Setup code to construct valid arguments for is_strictly_correct

ARGS = {
    'instance': {
        'label': 'SUPPORTS',
        'predicted_label': 'SUPPORTS',
        'predicted_evidence': [
            ['Wolfgang Niedecken', 'sentence', '1'],
            ['Some Page', 'cell', '1_19_0']
        ],
        'evidence': [
            [
                ['Wolfgang Niedecken', 'sentence', '1'],
                ['Some Page', 'cell', '1_19_0']
            ]
        ]
    }
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_sparse_csr Ran
ran on a synthesized input
Raldir/FEVEROUS/src/feverous/database/utils.py
code served (permissive licence) · get_code("71cbb05f2913e671")
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 numpy as np
import scipy.sparse as sp
import tempfile
import os

# Create a temporary sparse matrix file in NPZ format
temp_dir = tempfile.gettempdir()
temp_filename = os.path.join(temp_dir, "test_sparse_matrix.npz")

# Construct a small sparse CSR matrix
data = np.array([1, 2, 3, 4], dtype=np.float32)
indices = np.array([0, 2, 1, 3], dtype=np.int32)
indptr = np.array([0, 2, 4], dtype=np.int32)
shape = np.array([2, 4], dtype=np.int64)

# Create metadata
metadata = np.array([{"key": "value", "num": 42}], dtype=object)

# Save to NPZ file
np.savez(
    temp_filename,
    data=data,
    indices=indices,
    indptr=indptr,
    shape=shape,
    metadata=metadata
)

ARGS = {
    "filename": temp_filename
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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
normalize Ran
ran; our contract draft was wrong, not the code
Raldir/FEVEROUS/src/feverous/database/utils.py
code served (permissive licence) · get_code("b1c91c9e3a3b88e4")
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 unicodedata

# The normalize function takes a single parameter 'text' which should be a string
# The function uses unicodedata.normalize('NFD', text) to resolve unicode encodings

ARGS = {
    'text': 'café'  # A simple string with unicode characters that can be normalized
}
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
process_text Ran
ran on a synthesized input
Raldir/FEVEROUS/src/feverous/utils/wiki_element.py
code served (permissive licence) · get_code("52deef4a225ce9bf")
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)

ARGS = {
    "text": "  hello world  \n"
}
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
truncate_evidence Ran
ran on a synthesized input
Raldir/FEVEROUS/src/feverous/evaluation/feverous_scorer.py
code served (permissive licence) · get_code("3b3a46e54d1eb6ea")
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
import numpy as np

# Create a minimal instance dict with predicted_evidence structure
instance = {
    'predicted_evidence': [
        ['source1', 'cell', 'evidence_data_1'],
        ['source2', 'paragraph', 'evidence_data_2'],
        ['source3', 'item', 'evidence_data_3'],
        ['source4', 'paragraph', 'evidence_data_4'],
        ['source5', 'table_caption', 'evidence_data_5'],
        ['source6', 'paragraph', 'evidence_data_6'],
        ['source7', 'header_cell', 'evidence_data_7'],
        ['source8', 'paragraph', 'evidence_data_8'],
    ]
}

ARGS = {
    'instance': instance,
    'max_evidence': 2,
    'max_evidence_cell': 1
}
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 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_evidence_by_table Not yet run
unverified (harvested, no recorded run)
Raldir/FEVEROUS/src/feverous/utils/util.py
code served (permissive licence) · get_code("37d8a595f9c51936")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 4 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 1 the same month, 3 later) and 4 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
normalize 6 other papers
1 dated earlier, 4 later, 1 same year, order not determinable
1909.08041 (2019-09)
Revealing the Importance of Semantic Retrieval for Machine Reading at Scale
2021.acl-long.164 (2021 (year only))
BERTAC: Enhancing Transformer-based Language Models with Adversarially Pretrained Convolutional Neural Networks
2109.08535 (2021-09)
Simple Entity-Centric Questions Challenge Dense Retrievers
and 3 more not named here
hash 3 other papers
2 later, 1 same year, order not determinable
2021.acl-long.164 (2021 (year only))
BERTAC: Enhancing Transformer-based Language Models with Adversarially Pretrained Convolutional Neural Networks
2305.12744 (2023-05)
Fact-Checking Complex Claims with Program-Guided Reasoning
2505.22993 (2025-05)
Verify-in-the-Graph: Entity Disambiguation Enhancement for Complex Claim Verification with Interactive Graph Representation
is_correct_label 2 other papers
1 dated earlier, 1 the same month
1809.01479 (2018-09)
UKP-Athene: Multi-Sentence Textual Entailment for Claim Verification
2106.01191 (2021-06)
Topic-Aware Evidence Reasoning and Stance-Aware Aggregation for Fact Verification
load_sparse_csr 2 other papers
2 later
2305.12744 (2023-05)
Fact-Checking Complex Claims with Program-Guided Reasoning
2505.22993 (2025-05)
Verify-in-the-Graph: Entity Disambiguation Enhancement for Complex Claim Verification with Interactive Graph Representation

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

Fact verification has attracted a lot of attention in the machine learning and natural language processing communities, as it is one of the key methods for detecting misinformation. Existing large-scale benchmarks for this task have focused mostly on textual sources, i.e. unstructured information, and thus ignored the wealth of information available in structured formats, such as tables. In this paper we introduce a novel dataset and benchmark, Fact Extraction and VERification Over Unstructured and Structured information (FEVEROUS), which consists of 87,026 verified claims. Each claim is annotated with evidence in the form of sentences and/or cells from tables in Wikipedia, as well as a label indicating whether this evidence supports, refutes, or does not provide enough information to reach a verdict. Furthermore, we detail our efforts to track and minimize the biases present in the dataset and could be exploited by models, e.g. being able to predict the label without using evidence. Finally, we develop a baseline for verifying claims against text and tables which predicts both the correct evidence and verdict for 18% of the claims. * The author started working on this project whilst at Amazon. 35th Conference on Neural Information Processing Systems (NeurIPS 2021) Track on Datasets and Benchmarks.

For agents

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

get_harvested_code_for_paper("2106.05707")
get_code_for_paper("2106.05707")
have("2106.05707")

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

curl https://syntology.ai/api/ran/2106.05707.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.05707.svg)](https://syntology.ai/paper/2106.05707)

Connect an agent — have() is free.