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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 betweenstory

-0.59

PositiveButton

-0.47

RegressionTest

-0.47

UnknownFields

-0.47

soever

-0.46

GIVEREF

-0.45

Transkript

-0.43

ThroughAttribute

-0.42

 Administrativna

-0.41

 stolz

-0.41

POSITIVE LOGITS

 Roskov

1.41

 kasarigan

0.61

 nakalista

0.53

 للمعارف

0.49

Referensi

0.43

 châssis

0.40

 Италијани

0.40

ثيق

0.40

 informée

0.39

بوابة

0.39

Activations Density 0.000%

No Known Activations

This feature has no known activations.