INDEX

Explanations

news headlines or article titles that prompt the reader to "Read more."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Read," indicating sources or references for further information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

xon

-0.82

opard

-0.70

IDS

-0.70

IDA

-0.67

ounty

-0.65

TEXTURE

-0.65

OPER

-0.64

adish

-0.64

uay

-0.64

ascal

-0.63

POSITIVE LOGITS

 aloud

0.98

sburg

0.93

Write

0.86

iness

0.83

ahead

0.83

 Read

0.83

just

0.80

gon

0.80

ying

0.78

scl

0.78

Activations Density 0.014%

news headlines or article titles that prompt the reader to "Read more."

instances of the word "Read," indicating sources or references for further information

No Comments

No Known Activations