INDEX

Explanations

elements describing design features and attributes

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

 ejército

-0.36

 Ejército

-0.35

Mediabestanden

-0.34

另一种

-0.34

 barata

-0.33

 nôtre

-0.31

 Walkover

-0.31

 Imperio

-0.31

eclipse

-0.31

GestureDetector

-0.30

POSITIVE LOGITS

 autorytatywna

0.70

 صوتيه

0.68

setVerticalGroup

0.58

GIVEREF

0.56

Rhestr

0.54

onOptions

0.53

 שוליים

0.52

RenderAtEndOf

0.52

twimg

0.52

Dichloropropane

0.51

Activations Density 0.053%

elements describing design features and attributes

No Comments

No Known Activations