© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Neuronpedia

Natural Language

NEW Assistant AxisNEW Circuit TracerUPDATESteer SAE Evals ExportsAPI Community Blog Privacy & Terms Contact

QWEN2.5-7B-IT · 15-RESID-POST-AA · 41959 ｜ Neuronpedia

Home
Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
Qwen2.5-7B-IT
Resid Post - 131k
15-RESID-POST-AA
41959

INDEX

Explanations

provide

np_max-act · gemini-2.0-flash

New Auto-Interp

Top Features by Cosine Similarity

Configuration

andyrdt/saes-qwen2.5-7b-instruct/resid_post_layer_15/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.15.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

reference

-0.08

 Args

-0.07

NV

-0.07

.Compile

-0.07

 chase

-0.07

 vide

-0.07

ты

-0.07

watch

-0.06

 regain

-0.06

Message

-0.06

POSITIVE LOGITS

 Buddhism

0.07

땁

0.07

 arquivo

0.07

_PERSON

0.07

菊花

0.07

Grad

0.07

趺

0.07

 Дмитр

0.07

/css

0.07

מורה

0.07

Activations Density 0.005%

No Known Activations