INDEX

Explanations

numerical values indicating a decrease in something

oai_token-act-pair · gpt-3.5-turbo

phrases indicating reductions or decreases in various metrics or statistics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 ambitious

-0.76

 emot

-0.68

addy

-0.68

sty

-0.66

arthed

-0.66

 extraordinary

-0.66

 artic

-0.64

AFL

-0.64

 Ambassador

-0.63

hon

-0.63

POSITIVE LOGITS

 decrease

2.50

 decreases

2.11

 decreased

2.05

 decreasing

2.00

Decre

1.81

 Decre

1.69

 decline

1.30

 diminish

1.29

 lessen

1.21

 diminished

1.19

Activations Density 0.026%

numerical values indicating a decrease in something

phrases indicating reductions or decreases in various metrics or statistics

No Comments

No Known Activations