INDEX

Explanations

words indicating a strong preference or fondness

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

path for something

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

 שוליים

-0.54

stylers

-0.52

windowFixed

-0.47

 xăm

-0.46

lysninger

-0.46

UnitTesting

-0.45



-0.44



-0.43

 zoude

-0.43

mær

-0.42

POSITIVE LOGITS

ReusableCell

0.59

for

0.55

fjspx

0.52

flashdata

0.51

 nahilalakip

0.50

setColumns

0.47

makebox

0.46

__(/*!

0.46

Suitable

0.45

uitable

0.45

Activations Density 0.067%

words indicating a strong preference or fondness

path for something

No Comments

No Known Activations