INDEX

Explanations

phrases related to controversial or attention-grabbing actions or events

oai_token-act-pair · gpt-3.5-turbo

narratives related to pranks and political stunts

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

oj

-0.69

installed

-0.66

âĶ

-0.66

omes

-0.65

 listened

-0.63

 Notes

-0.62

readable

-0.62

 definitions

-0.61

afa

-0.61

 listens

-0.61

POSITIVE LOGITS

 stunt

3.99

 stunts

3.04

 prank

1.57

 gimmick

1.34

 ploy

1.22

 circus

1.18

 chore

1.09

 hoax

1.07

 feat

1.04

 trick

1.04

Activations Density 0.017%

phrases related to controversial or attention-grabbing actions or events

narratives related to pranks and political stunts

No Comments

No Known Activations