INDEX

Explanations

news headlines or article titles

oai_token-act-pair · gpt-3.5-turbo

instances of the word "READ" indicating calls to action or prompts 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 @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

phies

-0.68

angers

-0.67

phi

-0.66

phy

-0.66

aviour

-0.64

 Bret

-0.64

bows

-0.63

asketball

-0.63

amel

-0.62

phia

-0.62

POSITIVE LOGITS

 ALSO

1.01

ING

0.95

 MORE

0.94

MORE

0.89

LIST

0.86

INGS

0.85

 UPDATE

0.84

 DOWN

0.81

BOOK

0.80

ERS

0.80

Activations Density 0.010%

news headlines or article titles

instances of the word "READ" indicating calls to action or prompts for further information

No Comments

No Known Activations