INDEX

Explanations

details about events and their descriptions

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

photograph or image description

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

 autorytatywna

-0.79

SharedDtor

-0.49

Rptr

-0.49

addGap

-0.48

FunctionFlags

-0.47

Portail

-0.47

ⓧ

-0.46

kháu

-0.46

 moeite

-0.45

 Италијани

-0.45

POSITIVE LOGITS

 photograph

0.69

 photographs

0.63

photograph

0.61

 photo

0.58

 fotografi

0.56

 photographer

0.56

twimg

0.56

 photos

0.56

 photographed

0.56

 fotografico

0.55

Activations Density 0.091%

details about events and their descriptions

photograph or image description

No Comments

No Known Activations