INDEX

Explanations

phrases related to legal or official matters

oai_token-act-pair · gpt-3.5-turbo

phrases related to specific events, situations, or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 parity

-0.66

urg

-0.64

 maiden

-0.63

 compulsion

-0.62

 princip

-0.60

 referees

-0.59

Ĥ

-0.59

 altogether

-0.57

 associations

-0.57

 earnest

-0.57

POSITIVE LOGITS

 Photo

1.46

 PHOTO

1.38

reenshot

1.38

 Shutterstock

1.36

 Courtesy

1.36

 Photograph

1.34

 Photographer

1.32

Photo

1.29

 REUTERS

1.27

 Flickr

1.22

Activations Density 1.342%

phrases related to legal or official matters

phrases related to specific events, situations, or conditions

No Comments

No Known Activations