© 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

Home
Gemma-3-27B-IT
40-GEMMASCOPE-2-RES-16K
12556

INDEX

Explanations

Java

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

Negative Logits

Nim

0.40

 Lobo

0.40

Vitamin

0.39

Boa

0.38

iciens

0.38

 hermit

0.38

chakra

0.38

 pará

0.37

 createTask

0.37

Interested

0.37

POSITIVE LOGITS

 Java

0.89

Java

0.81

 JAVA

0.77

 java

0.65

java

0.63

JAVA

0.62

 javac

0.54

jav

0.49

jav

0.49

wsdl

0.46

Activations Density 0.057%

No Known Activations