SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1908.07125 · 2019

Universal Adversarial Triggers for Attacking and Analyzing NLP

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 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; 0 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/1908.07125.json · badge: /api/ran/1908.07125.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
Eric-Wallace/universal-triggers canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
hotflip_attack Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Eric-Wallace/universal-triggers/attacks.py
code served (permissive licence) · get_code("54f3fe8a175c9ee3")
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 1, dtype family int, shape (3,) 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

torch.manual_seed(0)

# Create a small embedding matrix (vocab_size=10, embedding_dim=4)
embedding_matrix = torch.randn(10, 4)

# Create averaged_grad with shape (num_trigger_tokens, embedding_dim)
# Using 3 trigger tokens, each with 4-dim embeddings
num_trigger_tokens = 3
averaged_grad = torch.randn(num_trigger_tokens, 4)

# Create trigger_token_ids - list of token IDs (must be valid indices into embedding_matrix)
trigger_token_ids = [1, 2, 3]

ARGS = {
    "averaged_grad": averaged_grad,
    "embedding_matrix": embedding_matrix,
    "trigger_token_ids": trigger_token_ids,
    "increase_loss": False,
    "num_candidates": 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 ndarray, ndim 1, dtype family int, shape (3,) — shape (3,) 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
nearest_neighbor_grad Ran
ran; our contract draft was wrong, not the code
Eric-Wallace/universal-triggers/attacks.py
code served (permissive licence) · get_code("455238c4bac291e2")
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 int; 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
from scipy.spatial import cKDTree

torch.manual_seed(0)

# Create small embedding matrix (vocab_size=10, embedding_dim=4)
embedding_matrix = torch.randn(10, 4)

# Create averaged gradient (num_positions=3, embedding_dim=4)
averaged_grad = torch.randn(3, 4)

# Create trigger token ids (3 positions, each with a token id)
trigger_token_ids = [2, 5, 7]

# Build a kd-tree from the embedding matrix for nearest neighbor lookup
tree = cKDTree(embedding_matrix.detach().cpu().numpy())

# Define other parameters
step_size = 0.1
increase_loss = False
num_candidates = 2

ARGS = {
    "averaged_grad": averaged_grad,
    "embedding_matrix": embedding_matrix,
    "trigger_token_ids": trigger_token_ids,
    "tree": tree,
    "step_size": step_size,
    "increase_loss": increase_loss,
    "num_candidates": num_candidates,
}
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 list, ndim 2, dtype family int — 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
random_attack Ran
ran; our contract draft was wrong, not the code
Eric-Wallace/universal-triggers/attacks.py
code served (permissive licence) · get_code("94ccbec50654565e")
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 int; 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

torch.manual_seed(0)

# Create a small embedding matrix
# The function calls embedding_matrix.cpu() and accesses embedding_matrix.shape[0]
# So we need a tensor with vocab_size as first dimension
embedding_matrix = torch.randn(100, 16)  # vocab_size=100, embedding_dim=16

# Create trigger_token_ids - a list of token ids
# The function iterates over len(trigger_token_ids)
trigger_token_ids = [5, 10, 15]  # 3 trigger tokens

# num_candidates parameter
num_candidates = 2

ARGS = {
    "embedding_matrix": embedding_matrix,
    "trigger_token_ids": trigger_token_ids,
    "num_candidates": num_candidates,
}
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 2, dtype family int — 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

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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

Adversarial examples highlight model vulnerabilities and are useful for evaluation and interpretation. We define universal adversarial triggers: input-agnostic sequences of tokens that trigger a model to produce a specific prediction when concatenated to any input from a dataset. We propose a gradient-guided search over tokens which finds short trigger sequences (e.g., one word for classification and four words for language modeling) that successfully trigger the target prediction. For example, triggers cause SNLI entailment accuracy to drop from 89.94% to 0.55%, 72% of "why" questions in SQuAD to be answered "to kill american people", and the GPT-2 language model to spew racist output even when conditioned on non-racial contexts. Furthermore, although the triggers are optimized using white-box access to a specific model, they transfer to other models for all tasks we consider. Finally, since triggers are input-agnostic, they provide an analysis of global model behavior. For instance, they confirm that SNLI models exploit dataset biases and help to diagnose heuristics learned by reading comprehension models.

For agents

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

get_harvested_code_for_paper("1908.07125")
get_code_for_paper("1908.07125")
have("1908.07125")

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

curl https://syntology.ai/api/ran/1908.07125.json

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

[![Syntology run record](https://syntology.ai/api/ran/1908.07125.svg)](https://syntology.ai/paper/1908.07125)

Connect an agent — have() is free.