INDEX

Explanations

punctuations and markers that often indicate a conversational or question format in text

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

enumi

-0.57

 henvisninger

-0.56

ViewImports

-0.47

 especiales

-0.47

 complémentaires

-0.46

 especiais

-0.46

 complementos

-0.45

 lainnya

-0.44

 muut

-0.43

 adicionais

-0.43

POSITIVE LOGITS

="@+

0.50

первых

0.48

Firstly

0.47

 Firstly

0.47

findpost

0.45

RTLU

0.42

PickerController

0.41

 ['./

0.39

 Primo

0.39

intios

0.38

Activations Density 0.089%

punctuations and markers that often indicate a conversational or question format in text

No Comments

No Known Activations