PulseAugur
EN
LIVE 17:56:08

OLMo 3 7B training reveals structured harmfulness directions

By PulseAugur Editorial · [1 sources] ·

Researchers have analyzed the development of harmfulness representations within the OLMo 3 7B model during its training process. They identified distinct but related linear activation directions for various harmfulness subcategories, observing that these directions evolve and stabilize over time. The study found that in-distribution evaluations can be misleading, emphasizing the need for out-of-distribution testing, and demonstrated that late-stage training directions can effectively steer the model's behavior. AI

IMPACT Reveals insights into how harmful concepts are represented and evolve during LLM training, potentially informing future safety research.

RANK_REASON Technical report detailing methodology and findings on model training dynamics and harmfulness representations. [lever_c_demoted from research: ic=1 ai=1.0]

Read on LessWrong (AI tag) →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

OLMo 3 7B training reveals structured harmfulness directions

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Technical report detailing methodology and findings on model training dynamics and harmfulness representations. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
120 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

Full methodology in our editorial standards.

COVERAGE [1]

  1. LessWrong (AI tag) TIER_1 English(EN) · Daniele Pace ·

    Harmfulness Directions in OLMo

    <img alt="pca_centroids_animation.gif" src="https://res.cloudinary.com/lesswrong-2-0/image/upload/v1778859923/lexical_client_uploads/kbq9rgmrhedzrnngtmkn.gif" /><h1><span>Introduction</span></h1><p><span>This work was conducted as part of the MARS 4.0 program, supervised by Loren…

RELATED ENTITIES

RELATED TOPICS