INDEX

Explanations

words and phrases related to accidental actions or occurrences

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

område

-0.56

 ModelExpression

-0.55

 Infórmanos

-0.54

了嗎

-0.52

 cielos

-0.52

jaqueta

-0.51

 kasarigan

-0.50

\{\\

-0.50

 Theſe

-0.49

 juſ

-0.49

POSITIVE LOGITS

 accidental

0.87

 unintentional

0.86

 accidentally

0.86

 unintentionally

0.83

Accidental

0.81

 Accidental

0.80

 inadvert

0.79

 inadvertently

0.77

accident

0.67

 unintended

0.65

Activations Density 0.014%

words and phrases related to accidental actions or occurrences

No Comments

No Known Activations