INDEX

Explanations

adjectives indicating extremity or intensity

oai_token-act-pair · gpt-3.5-turbo

superlative adjectives that emphasize negativity or controversy

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 pione

-0.81

StreamerBot

-0.80

ActionCode

-0.79

č

-0.79

Ė

-0.79

ė

-0.78

ē

-0.78

 RandomRedditor

-0.78

ÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤ

-0.78

POSITIVE LOGITS

 imaginable

1.31

 aspect

1.07

 conceivable

1.03

of

1.02

 aspects

0.96

 thing

0.95

 manifestation

0.94

 element

0.87

 incarnation

0.85

 ever

0.85

Activations Density 0.150%

adjectives indicating extremity or intensity

superlative adjectives that emphasize negativity or controversy

No Comments

No Known Activations