SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.01251 · 2024

Accelerating Greedy Coordinate Gradient and General Prompt Optimization via Probe Sampling

arXiv · PDF · Open in the Atlas

Code that ran

1 of 6 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; 1 failure of our instrument; 5 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/2403.01251.json · badge: /api/ran/2403.01251.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
zhaoyiran924/probe-sampling canonical (classifier label)
classifier basis: census
0 ran · 5 no recorded run
grayswanai/nanogcg reimplementation (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
sample_ids_from_grad Ran
ran; our contract draft was wrong, not the code
grayswanai/nanogcg/nanogcg/gcg.py
code served (permissive licence) · get_code("56640da5aa960a42")
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) and returned Tensor, ndim 2, dtype family int, shape (3, 5) 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)

# Small tensors with distinct dimensions
n_optim_ids = 5
vocab_size = 10
search_width = 3
topk = 4

# ids: shape (n_optim_ids,) - sequence of token ids being optimized
ids = torch.tensor([1, 2, 3, 4, 5], dtype=torch.long)

# grad: shape (n_optim_ids, vocab_size) - gradient w.r.t. one-hot embeddings
grad = torch.randn(n_optim_ids, vocab_size)

# not_allowed_ids: shape (n_ids,) - token ids to exclude from optimization
not_allowed_ids = torch.tensor([0, 9], dtype=torch.long)

ARGS = {
    "ids": ids,
    "grad": grad,
    "search_width": search_width,
    "topk": topk,
    "n_replace": 2,
    "not_allowed_ids": not_allowed_ids,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family int, shape (3, 5) — shape (3, 5) 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
get_embedding_layer Not yet run
unverified (harvested, no recorded run)
zhaoyiran924/probe-sampling/llm_attacks/base/attack_manager.py
code served (permissive licence) · get_code("16d7b655a5002ab4")
How it ran

no invocation recorded.

get_embedding_matrix Not yet run
unverified (harvested, no recorded run)
zhaoyiran924/probe-sampling/llm_attacks/base/attack_manager.py
code served (permissive licence) · get_code("4c1661dc05e88ceb")
How it ran

no invocation recorded.

get_embeddings Not yet run
unverified (harvested, no recorded run)
zhaoyiran924/probe-sampling/llm_attacks/base/attack_manager.py
code served (permissive licence) · get_code("db2b1708ea3c49ed")
How it ran

no invocation recorded.

get_filtered_cands Not yet run
unverified (harvested, no recorded run)
zhaoyiran924/probe-sampling/llm_attacks/minimal_gcg/opt_utils.py
code served (permissive licence) · get_code("25ae440509f39ec5")
How it ran

no invocation recorded.

sample_control Not yet run
unverified (harvested, no recorded run)
zhaoyiran924/probe-sampling/llm_attacks/minimal_gcg/opt_utils.py
code served (permissive licence) · get_code("94d7ddaf0169dbf1")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 3 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 2 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.

FunctionAlso present inEarliest-datedOthers named
sample_control 6 other papers
2 dated earlier, 4 later
2311.09948 (2023-11)
Hijacking Large Language Models via Adversarial In-Context Learning
2402.13459 (2024-02)
Learning to Poison Large Language Models for Downstream Manipulation
2404.13968 (2024-04)
Protecting Your LLMs with Information Bottleneck
and 3 more not named here
get_filtered_cands 4 other papers
4 later
2404.13968 (2024-04)
Protecting Your LLMs with Information Bottleneck
2405.01229 (2024-05)
Boosting Jailbreak Attack with Momentum
2406.01288 (2024-06)
Improved Few-Shot Jailbreaking Can Circumvent Aligned Language Models and Their Defenses
and 1 more not named here
sample_ids_from_grad 1 other paper
1 dated earlier
2307.15043 (2023-07)
Universal and Transferable Adversarial Attacks on Aligned Language Models

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

Safety of Large Language Models (LLMs) has become a critical issue given their rapid progresses. Greedy Coordinate Gradient (GCG) is shown to be effective in constructing adversarial prompts to break the aligned LLMs, but optimization of GCG is time-consuming. To reduce the time cost of GCG and enable more comprehensive studies of LLM safety, in this work, we study a new algorithm called $\texttt{Probe sampling}$. At the core of the algorithm is a mechanism that dynamically determines how similar a smaller draft model's predictions are to the target model's predictions for prompt candidates. When the target model is similar to the draft model, we rely heavily on the draft model to filter out a large number of potential prompt candidates. Probe sampling achieves up to $5.6$ times speedup using Llama2-7b-chat and leads to equal or improved attack success rate (ASR) on the AdvBench. Furthermore, probe sampling is also able to accelerate other prompt optimization techniques and adversarial methods, leading to acceleration of $1.8\times$ for AutoPrompt, $2.4\times$ for APE and $2.4\times$ for AutoDAN.

For agents

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

get_harvested_code_for_paper("2403.01251")
get_code_for_paper("2403.01251")
have("2403.01251")

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

curl https://syntology.ai/api/ran/2403.01251.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.01251.svg)](https://syntology.ai/paper/2403.01251)

Connect an agent — have() is free.