The reviewed record of science sign in
Pith

arxiv: 2108.07790 · v3 · pith:URRXZ7SK · submitted 2021-08-04 · cs.CL · cs.LG

Mitigating harm in language models with conditional-likelihood filtration

Helen Ngo , Cooper Raterink , Jo\~ao G.M. Ara\'ujo , Ivan Zhang , Carol Chen , Adrien Morisot , Nicholas Frosst This is my paper

Reviewed by Pithpith:URRXZ7SKopen to challenge →

classification cs.CL cs.LG
keywords languagemodelsharmfulusedbenchmarksdatasetdatasetsmodeling
0
0 comments X
read the original abstract

Language models trained on large-scale unfiltered datasets curated from the open web acquire systemic biases, prejudices, and harmful views from their training data. We present a methodology for programmatically identifying and removing harmful text from web-scale datasets. A pretrained language model is used to calculate the log-likelihood of researcher-written trigger phrases conditioned on a specific document, which is used to identify and filter documents from the dataset. We demonstrate that models trained on this filtered dataset exhibit lower propensity to generate harmful text, with a marginal decrease in performance on standard language modeling benchmarks compared to unfiltered baselines. We provide a partial explanation for this performance gap by surfacing examples of hate speech and other undesirable content from standard language modeling benchmarks. Finally, we discuss the generalization of this method and how trigger phrases which reflect specific values can be used by researchers to build language models which are more closely aligned with their values.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Low-Resource Languages Jailbreak GPT-4

    cs.CL 2023-10 conditional novelty 6.0

    Translating unsafe inputs to low-resource languages jailbreaks GPT-4 at rates on par with or exceeding state-of-the-art attacks.

  2. A General Language Assistant as a Laboratory for Alignment

    cs.CL 2021-12 conditional novelty 6.0

    Ranked preference modeling outperforms imitation learning for language model alignment and scales more favorably with model size.

  3. Harder to Defend: Towards Chinese Toxicity Attacks via Implicit Enhancement and Obfuscation Rewriting

    cs.CL 2026-05 unverdicted novelty 5.0

    CITA generates Chinese implicit toxicity samples that cause 69.48% average missed detection across seven tested detectors while preserving harmfulness, and the same data improves robustness when used to fine-tune a CI...

  4. Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment

    cs.AI 2023-08 accept novelty 5.0

    Survey organizes LLM trustworthiness into seven categories and 29 sub-categories, measures eight sub-categories on popular models, and finds that more aligned models generally score higher but with varying effectiveness.

  5. Using DeepSpeed and Megatron to Train Megatron-Turing NLG 530B, A Large-Scale Generative Language Model

    cs.CL 2022-01 unverdicted novelty 5.0

    Trained the largest monolithic 530B-parameter transformer language model to date and reported new state-of-the-art zero- and few-shot results on multiple NLP benchmarks.