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

CMR

-0.35

AccessFile

-0.35

 yogur

-0.35

 tagging

-0.33

ssen

-0.33

aff

-0.33

 queries

-0.33

比

-0.33

ragos

-0.33

ro

-0.32

POSITIVE LOGITS

ној

2.06

ској

1.33

 Meksiku

0.98

 држа

0.96

!*\

0.81

ніципа

0.76

 CanadaChoose

0.76

ніципалі

0.72

 autorytatywna

0.68

 disambiguazione

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.