INDEX

Explanations

phrases indicating a count or categorization

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 initComponents

-0.54

AutoScaleMode

-0.54

viewDidLoad

-0.54

AutoModerator

-0.53

 Normdatei

-0.53

Tembelea

-0.52

viewtopic

-0.52

MessageTagHelper

-0.48

amentals

-0.48

 pictureBox

-0.47

POSITIVE LOGITS

 terdiri

0.41

ufficio

0.40

 waaronder

0.40

 incluidas

0.40

 jonka

0.39

 diantaranya

0.38

 incluidos

0.37

 vklju

0.36

 Hälfte

0.36

 dotyczy

0.35

Activations Density 0.028%

phrases indicating a count or categorization

No Comments

No Known Activations