INDEX

Explanations

phrases indicating legal or compliance-related issues

oai_token-act-pair · gpt-4o-mini Triggered by @bot

sufficient to apply

np_acts-logits-general · gemini-2.5-flash-lite

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

ViewFeatures

-0.45

surate

-0.45

TabIndex

-0.41

ContentAlignment

-0.41

oplayer

-0.41

 Brist

-0.40

ärä

-0.39

 StatefulWidget

-0.38

Cyfarwyddwr

-0.38

rlrl

-0.37

POSITIVE LOGITS

DockStyle

0.56

IsMutable

0.45

锈钢

0.43

 GenerationType

0.40

 způso

0.39

}{@

0.39

CPtr

0.38

quelize

0.37

 excepción

0.37

 čern

0.36

Activations Density 0.143%

phrases indicating legal or compliance-related issues

sufficient to apply

No Comments

No Known Activations