INDEX

Explanations

numbers

oai_token-act-pair · gpt-3.5-turbo

numerical statistics or percentages in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

LAN

-0.73

izont

-0.70

Broad

-0.69

aucus

-0.69

achu

-0.69

Activity

-0.68

redit

-0.68

Cra

-0.68

helps

-0.68

avior

-0.68

POSITIVE LOGITS

 thousand

1.14

 percent

1.03

een

0.99

 cents

0.94

teen

0.92

 hundred

0.91

eenth

0.89

 million

0.84

 pounds

0.84

 dollars

0.79

Activations Density 0.049%

numbers

numerical statistics or percentages in the text

No Comments

No Known Activations