© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Neuronpedia

APIAssistant AxisNEW Circuit TracerNEW Steer SAE Evals Exports Community Blog Privacy & Terms Contact

LLAMA3.1-8B-IT · 3-RESID-POST-AA · 95303 ｜ Neuronpedia

Home
Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
Llama3.1-8B-IT (Instruct)
Resid Post - 131k
3-RESID-POST-AA
95303

INDEX

Explanations

-resistant

np_max-act · gemini-2.0-flash

New Auto-Interp

Top Features by Cosine Similarity

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_3/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.3.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

Show PlotsShow ExplanationShow ActivationsShow Test FieldShow SteerShow Link

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cimiento

-0.07

 oldu

-0.07

-0.06

ngh

-0.06

bil

-0.06

 Tween

-0.06

 기간

-0.06

Whitespace

-0.06

 continu

-0.06

(where

-0.06

POSITIVE LOGITS

 дів

0.07

#$

0.06

.Help

0.06

 hubby

0.06

margin

0.06

	Default

0.06

INDOW

0.06

 talents

0.06

 complicated

0.06

�

0.06

Activations Density 0.003%

No Known Activations