INDEX

Explanations

words related to repetition of actions or events

oai_token-act-pair · gpt-3.5-turbo

instances of the prefix "re" indicating actions or concepts of returning or repeating

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Phar

-0.73

DragonMagazine

-0.72

 Nieto

-0.70

INESS

-0.69

 Sturgeon

-0.65

 Everywhere

-0.64

 Institution

-0.64

 Icelandic

-0.64

 Ideal

-0.63

lihood

-0.62

POSITIVE LOGITS

becca

1.13

integ

1.03

ordered

1.03

authorized

1.00

interpret

0.99

clamation

0.98

uters

0.96

apers

0.96

forestation

0.95

charging

0.93

Activations Density 0.033%

words related to repetition of actions or events

instances of the prefix "re" indicating actions or concepts of returning or repeating

No Comments

No Known Activations