© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Neuronpedia

Jacobian LensNEW

Natural Language

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

OLMO-3-1125-32B · 32-RES-BATCHTOPK-131K · 5688 ｜ Neuronpedia

Home
Olmo-3-1125-32B
32-RES-BATCHTOPK-131K
5688

INDEX

Explanations

killing people

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

New Auto-Interp

Top Features by Cosine Similarity

Configuration

bcywinski/Olmo-3-32B-Base-SAE/saes_allenai_Olmo-3-1125-32B_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

XS

-0.10

 mour

-0.10

Roh

-0.09

-assets

-0.09

 rods

-0.09

 presses

-0.09

NEL

-0.09

dp

-0.08

lug

-0.08

isko

-0.08

POSITIVE LOGITS

 innocent

0.14

 people

0.14

 guards

0.13

 dozens

0.13

 unlucky

0.12

 Guards

0.12

äºº

0.12

 unfortunate

0.11

 lives

0.11

 innoc

0.11

Activations Density 0.191%

No Known Activations