INDEX

Explanations

references to video games and their narratives involving conflict and characters

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

descriptive words followed by nouns

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

 conmigo

-0.39

 ordinaires

-0.37

 mío

-0.34

 fija

-0.33

 Suiza

-0.33

 also

-0.32

 Egipto

-0.32

 very

-0.31

 rigtig

-0.31

 amarillas

-0.30

POSITIVE LOGITS

 Infórmanos

0.93

 estekak

0.87

+#+#

0.86

ंदीखरीदारी

0.84

 queſta

0.82

sizeCache

0.80

ThroughAttribute

0.79

$_(

0.79

LookAnd

0.78

脚注の使い方

0.78

Activations Density 0.048%

references to video games and their narratives involving conflict and characters

descriptive words followed by nouns

No Comments

No Known Activations