INDEX

Explanations

statements about the usefulness or significance of various concepts or findings

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

 Pretty

-0.38

Alt

-0.37

GI

-0.37

 Else

-0.36

 pumpkin

-0.35

 colors

-0.35

 Military

-0.35

Alt

-0.34

Else

-0.34

Ex

-0.33

POSITIVE LOGITS

SequentialGroup

0.62

 faptul

0.59

 Gedanke

0.56

 mahdol

0.56

 rozho

0.55

 Wikiseite

0.54

colgante

0.54

 Tatsache

0.52

Sklici

0.52

stateProvider

0.52

Activations Density 0.092%

statements about the usefulness or significance of various concepts or findings

No Comments

No Known Activations