© 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 · 39519 ｜ Neuronpedia

Home
Qwen3-32B
32-RESID-BATCHTOPK-65K
39519

INDEX

Explanations

broken noun phrases

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

 Crimes

-0.10

å£«

-0.09

å®¡

-0.09

Bubble

-0.09

rog

-0.08

 ÑģÐ°Ð¼ÑĭÐµ

-0.08

ç¼¢

-0.08

/story

-0.08

eral

-0.08

 ÑģÐ°Ð¼ÑĭÑħ

-0.08

POSITIVE LOGITS

ness

0.16

ated

0.12

ly

0.11

ä¸įåłª

0.11

ating

0.10

NESS

0.10

lier

0.10

 minded

0.09

ï¸

0.09

ides

0.09

Activations Density 0.019%

No Known Activations