SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2011.10369 · 2020

ONION: A Simple and Effective Defense Against Textual Backdoor Attacks

arXiv · PDF · Open in the Atlas

Code that ran

3 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 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/2011.10369.json · badge: /api/ran/2011.10369.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
thunlp/ONION canonical (classifier label)
classifier basis: classifier_2025
3 ran · 1 no recorded run
FunctionStatusWhere it lives
filter_sent Ran
ran; our contract draft was wrong, not the code
thunlp/ONION/experiments/test_defense.py
code served (permissive licence) · get_code("0a3034959f19105b")
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)

# split_sent is a list of words (strings)
split_sent = ["hello", "world", "foo", "bar", "baz"]

# pos is an integer index
pos = 2

ARGS = {
    "split_sent": split_sent,
    "pos": pos
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_all_data Ran
ran; our contract draft was wrong, not the code
thunlp/ONION/experiments/run_poison_bert.py
code served (permissive licence) · get_code("0762630368cd3261")
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 3, 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 os
import tempfile
import pandas as pd

# Create temporary directory with test TSV files
temp_dir = tempfile.mkdtemp()

# Create train.tsv
train_data = [
    ["This is a positive sentence", "1"],
    ["This is a negative sentence", "0"],
]
train_df = pd.DataFrame(train_data)
train_path = os.path.join(temp_dir, 'train.tsv')
train_df.to_csv(train_path, sep='\t', header=False, index=False)

# Create dev.tsv
dev_data = [
    ["This is another positive example", "1"],
    ["This is another negative example", "0"],
]
dev_df = pd.DataFrame(dev_data)
dev_path = os.path.join(temp_dir, 'dev.tsv')
dev_df.to_csv(dev_path, sep='\t', header=False, index=False)

# Create test.tsv
test_data = [
    ["Final positive test sentence", "1"],
    ["Final negative test sentence", "0"],
]
test_df = pd.DataFrame(test_data)
test_path = os.path.join(temp_dir, 'test.tsv')
test_df.to_csv(test_path, sep='\t', header=False, index=False)

ARGS = {
    'base_path': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 3, 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
read_data Ran
ran; our contract draft was wrong, not the code
thunlp/ONION/experiments/run_poison_bert.py
code served (permissive licence) · get_code("213b3de0b016d2f7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import tempfile
import os

# Create a temporary TSV file with test data
temp_dir = tempfile.mkdtemp()
file_path = os.path.join(temp_dir, "test_data.tsv")

# Write sample data to the TSV file
with open(file_path, 'w') as f:
    f.write("This is a sample sentence\t0\n")
    f.write("This is another sentence\t1\n")
    f.write("Yet another test sentence\t0\n")

ARGS = {
    "file_path": file_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
evaluaion Not yet run
unverified (harvested, no recorded run)
thunlp/ONION/experiments/run_poison_bert.py
code served (permissive licence) · get_code("31ff894ab3f7a745")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 3 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 3 later) and 1 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
evaluaion 1 other paper
1 later
2407.19845 (2024-07)
BackdoorBench: A Comprehensive Benchmark and Analysis of Backdoor Learning
get_all_data 1 other paper
1 later
2407.19845 (2024-07)
BackdoorBench: A Comprehensive Benchmark and Analysis of Backdoor Learning
read_data 1 other paper
1 later
2407.19845 (2024-07)
BackdoorBench: A Comprehensive Benchmark and Analysis of Backdoor Learning

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

Backdoor attacks are a kind of emergent training-time threat to deep neural networks (DNNs). They can manipulate the output of DNNs and possess high insidiousness. In the field of natural language processing, some attack methods have been proposed and achieve very high attack success rates on multiple popular models. Nevertheless, there are few studies on defending against textual backdoor attacks. In this paper, we propose a simple and effective textual backdoor defense named ONION, which is based on outlier word detection and, to the best of our knowledge, is the first method that can handle all the textual backdoor attack situations. Experiments demonstrate the effectiveness of our model in defending BiLSTM and BERT against five different backdoor attacks. All the code and data of this paper can be obtained at https://github.com/thunlp/ONION.

For agents

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

get_harvested_code_for_paper("2011.10369")
get_code_for_paper("2011.10369")
have("2011.10369")

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

curl https://syntology.ai/api/ran/2011.10369.json

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

[![Syntology run record](https://syntology.ai/api/ran/2011.10369.svg)](https://syntology.ai/paper/2011.10369)

Connect an agent — have() is free.