© 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 · 2190 ｜ Neuronpedia

Home
Qwen3-32B
32-RESID-BATCHTOPK-65K
2190

INDEX

Explanations

hit, hits, hit-to-lead

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

chor

-0.10

ainter

-0.10

ç¯ģ

-0.09

è²Į

-0.09

 Ð²Ð¾ÑĤ

-0.09

oes

-0.09

OTA

-0.09

riba

-0.09

lÄ±

-0.09

ities

-0.09

POSITIVE LOGITS

ting

0.14

TING

0.14

tings

0.14

ween

0.13

backs

0.12

iting

0.11

ters

0.11

æ¦ľ

0.11

uated

0.11

../

0.10

Activations Density 0.024%

No Known Activations