INDEX

Explanations

the word "by."

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

describing action or method

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

the

-0.47

-0.38

 deuil

-0.38

 męska

-0.36

 zimowa

-0.36

 spørs

-0.35

"{}

-0.34

 rodz

-0.34

 czarna

-0.33

有什麼

-0.33

POSITIVE LOGITS

 betweenstory

0.72

 nahilalakip

0.70

ьаж

0.69

ConstraintMaker

0.69

awtextra

0.68

complexContent

0.67

jsdelivr

0.66

 мәкал

0.65

 AssemblyCulture

0.65

WebElementEntity

0.63

Activations Density 0.067%

the word "by."

describing action or method

No Comments

No Known Activations