INDEX

Explanations

sequences of whitespace and formatting characters

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

 resourceCulture

-0.54

 enfans

-0.50

Legături

-0.50

 magasiner

-0.49

 autorytatywna

-0.49

 ainfi

-0.48

 rallado

-0.47

 vectorielle

-0.47



-0.47

IContainer

-0.46

POSITIVE LOGITS

0.56

 term

0.52

cot

0.52

0.51

{~

0.50

 usual

0.49

 Term

0.49

0.48

Term

0.48

Activations Density 0.002%

sequences of whitespace and formatting characters

No Comments

No Known Activations