INDEX

Explanations

the word "creep" or variations of it

oai_token-act-pair · gpt-3.5-turbo

variations of the word "creep."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ctions

-0.89

ctic

-0.82

nas

-0.76

sol

-0.75

ction

-0.75

fman

-0.74

lain

-0.71

pled

-0.70

HCR

-0.69

ufact

-0.69

POSITIVE LOGITS

 creep

1.36

 Creep

1.24

 creeps

1.18

 creeping

0.90

 crept

0.87

 craw

0.81

 crawl

0.80

iness

0.80

 aggro

0.73

stead

0.72

Activations Density 0.006%

the word "creep" or variations of it

variations of the word "creep."

No Comments

No Known Activations