© 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

GEMMA-3-27B-IT · 40-GEMMASCOPE-2-RES-65K · 3553 ｜ Neuronpedia

Home
Gemma-3-27B-IT
40-GEMMASCOPE-2-RES-65K
3553

INDEX

Explanations

purpose homework claim

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

New Auto-Interp

Top Features by Cosine Similarity

Configuration

google/gemma-scope-2-27b-it/resid_post/layer_40_width_65k_l0_medium

Prompts (Dashboard)

238,145 prompts, 512 tokens each

Dataset (Dashboard)

lmsys + oasst1

No Configuration Found

Embeds

Show PlotsShow ExplanationShow ActivationsShow Test FieldShow SteerShow Link

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

0.35

↵

0.34

 GameObject

0.34

 oder

0.34

も

0.33

ی

0.31

 things

0.31

0.30

（

0.30

 features

0.30

POSITIVE LOGITS

휩

0.41

ѕ

0.40

 แรก

0.39

:,

0.39

Allows

0.38

 обеспечить

0.38

𝙎

0.38

 ця

0.37

 сердце

0.37

 тепер

0.37

Activations Density 0.198%

No Known Activations