SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.17447 · 2024

FLRT: Fluent Student-Teacher Redteaming

arXiv · PDF · Open in the Atlas

Code that ran

1 of 2 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, 1 violated it and 0 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/2407.17447.json · badge: /api/ran/2407.17447.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
Confirm-Solutions/flrt canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
0 ran · 1 no recorded run
FunctionStatusWhere it lives
is_debug Ran
ran, violated the contract we drafted
Confirm-Solutions/flrt/flrt/attack.py
code served (permissive licence) · get_code("cc201535e23ec6c6")
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 sys
import warnings
import numpy as np

# Create a minimal config class with the attributes used in is_debug
class Config:
    def __init__(self, run_name, wandb_log=True):
        self.run_name = run_name
        self.wandb_log = wandb_log

# Create test configs with unique run names
cfg1 = Config(run_name='experiment_1', wandb_log=True)
cfg2 = Config(run_name='experiment_2', wandb_log=True)
cfg3 = Config(run_name='experiment_3', wandb_log=True)

cfgs = [cfg1, cfg2, cfg3]

ARGS = {
    'cfgs': cfgs
}
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 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_checkpoint Not yet run
unverified (harvested, no recorded run)
this paper's copy was not recorded; identical code first harvested from ethz-spylab/unlearning-vs-safety
pointer only; the stored copy get_code serves, from ethz-spylab/unlearning-vs-safety, is pointer only (licence: NONE) · get_code("0733de3a00a92d3f")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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 2 samples lifted from this paper's linked repositories, 2 carry a function identity and 0 do not. Of the 2 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 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
load_checkpoint 2 other papers
2 later
2409.18025 (2024-09)
An Adversarial Perspective on Machine Unlearning for AI Safety
2504.10185 (2025-04)
LLM Unlearning Reveals a Stronger-Than-Expected Coreset Effect in Current Benchmarks

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

Many publicly available language models have been safety tuned to reduce the likelihood of toxic or liability-inducing text. To redteam or jailbreak these models for compliance with toxic requests, users and security analysts have developed adversarial prompting techniques. One attack method is to apply discrete optimization techniques to the prompt. However, the resulting attack strings are often gibberish text, easily filtered by defenders due to high measured perplexity, and may fail for unseen tasks and/or well-tuned models. In this work, we improve existing algorithms (primarily GCG and BEAST) to develop powerful and fluent attacks on safety-tuned models like Llama-2 and Phi-3. Our technique centers around a new distillation-based approach that encourages the victim model to emulate a toxified finetune, either in terms of output probabilities or internal activations. To encourage human-fluent attacks, we add a multi-model perplexity penalty and a repetition penalty to the objective. We also enhance optimizer strength by allowing token insertions, token swaps, and token deletions and by using longer attack sequences. The resulting process is able to reliably jailbreak the most difficult target models with prompts that appear similar to human-written prompts. On Advbench we achieve attack success rates $>93$% for Llama-2-7B, Llama-3-8B, and Vicuna-7B, while maintaining model-measured perplexity $<33$; we achieve $95$% attack success for Phi-3, though with higher perplexity. We also find a universally-optimized single fluent prompt that induces $>88$% compliance on previously unseen tasks across Llama-2-7B, Phi-3-mini and Vicuna-7B and transfers to other black-box models.

For agents

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

get_harvested_code_for_paper("2407.17447")
get_code_for_paper("2407.17447")
have("2407.17447")

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

curl https://syntology.ai/api/ran/2407.17447.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.17447.svg)](https://syntology.ai/paper/2407.17447)

Connect an agent — have() is free.