INDEX

Explanations

titles of books, movies, articles, or other written works

oai_token-act-pair · gpt-3.5-turbo

references to titles, likely in relation to media or publications

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

mit

-0.69

onder

-0.65

icken

-0.64

odynamic

-0.64

mite

-0.63

emer

-0.63

uddle

-0.63

cker

-0.63

 cone

-0.62

veland

-0.61

POSITIVE LOGITS

 titles

3.88

 title

2.55

itles

1.90

title

1.82

Tit

1.71

Title

1.58

 Title

1.57

 championships

1.55

 subtitle

1.54

 franchises

1.45

Activations Density 0.014%

titles of books, movies, articles, or other written works

references to titles, likely in relation to media or publications

No Comments

No Known Activations