INDEX

Explanations

patterns related to votes and polling results

oai_token-act-pair · gpt-3.5-turbo

patterns of punctuation and ellipses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

board

-0.73

pher

-0.71

psons

-0.71

 Fulton

-0.70

boarding

-0.70

hof

-0.68

atten

-0.66

imb

-0.66

houses

-0.65

bat

-0.62

POSITIVE LOGITS

.........

1.32

.......

1.22

......

1.12

.....

1.11

?????

1.11

....

0.98

!!!!!

0.96

ãħĭ

0.95

..

0.87

???

0.86

Activations Density 0.009%

patterns related to votes and polling results

patterns of punctuation and ellipses

No Comments

No Known Activations