INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

GetMapping

-0.43

 miglia

-0.37

 podí

-0.35

IsNullOrEmpty

-0.34

 magát

-0.34

 dokun

-0.33

 noastră

-0.32

 människor

-0.31

ComponentScan

-0.31

かけ

-0.31

POSITIVE LOGITS

 revised

0.90

revised

0.78

 revision

0.76

 estekak

0.69

 Revised

0.68

 revise

0.66

Revised

0.66

 Revision

0.65

 revisions

0.64

interopRequire

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.