© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Neuronpedia

Jacobian LensNEW

Natural Language

NEW Assistant AxisNEW Circuit TracerUPDATESteer SAE Evals Exports interp-engineNEWAPI Community Blog Privacy & Terms Contact

QWEN3-32B · 32-RESID-BATCHTOPK-65K · 31056 ｜ Neuronpedia

Home
Qwen3-32B
32-RESID-BATCHTOPK-65K
31056

INDEX

Explanations

ee words followed by v or vi

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Configuration

adamkarvonen/qwen3-32b-saes/saes_Qwen_Qwen3-32B_batch_top_k/resid_post_layer_32

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

No Configuration Found

Embeds

Show PlotsShow ExplanationShow ActivationsShow Test FieldShow SteerShow Link

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ness

-0.10

à¸¢

-0.10

itecture

-0.10

nya

-0.09

ened

-0.09

icket

-0.09

ç»

-0.09

é²Ķ

-0.09

reeNode

-0.09

sWith

-0.09

POSITIVE LOGITS

zel

0.10

ption

0.10

otle

0.10

rott

0.10

ì¦Ī

0.10

quire

0.10

Ø§

0.09

0.09

ÙĤØª

0.09

quence

0.09

Activations Density 0.045%

No Known Activations