© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Jacobian LensNEW

Natural Language

MAY 26 Assistant AxisJAN 26 Circuit TracerAPR 26Steer SAE Evals Exports interp-engineNEW Community Blog Privacy & Terms Contact

GEMMA-3-27B-IT · 40-GEMMASCOPE-2-RES-16K · 8085 ｜ Neuronpedia

Home
Gemma-3-27B-IT
40-GEMMASCOPE-2-RES-16K
8085

INDEX

Explanations

how or what

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_16k_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

ப்படுகின்றன

0.30

 লিখিয়াছেন

0.29

ö

0.29

 ethical

0.28

మిక

0.28

0.27

ethical

0.27

í

0.27

eds

0.27

AP

0.26

POSITIVE LOGITS

how

0.35

 cómo

0.34

 ничек

0.31

 boundedness

0.31

 كيفية

0.31

 щодо

0.31

how

0.30

 কীভাবে

0.30

憊

0.30

 whats

0.29

Activations Density 0.193%

No Known Activations