INDEX

Explanations

interrogative phrases, specifically questions starting with "What," "How," and "Who."

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

what is the

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

 оригіналу

-0.71

 queſta

-0.68

setVerticalGroup

-0.67

bootstrapcdn

-0.67

tanleria

-0.66

 autorytatywna

-0.66

RTLR

-0.65

 convertView

-0.65

 تضيفلها

-0.65

 Winaray

-0.64

POSITIVE LOGITS

who

0.38

 fastest

0.34

有多少

0.33

’

0.33

ブリッド

0.32

 Horowitz

0.32

 about

0.31

how

0.31

 civilización

0.31

 eléctrico

0.31

Activations Density 0.077%

interrogative phrases, specifically questions starting with "What," "How," and "Who."

what is the

No Comments

No Known Activations