INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 herb

-0.73

swer

-0.71

 guesses

-0.66

swers

-0.64

inctions

-0.64

 ecstasy

-0.63

answered

-0.63

 sear

-0.62

lah

-0.62

 ascended

-0.62

POSITIVE LOGITS

utor

0.80

ilateral

0.68

ukong

0.68

station

0.66

riter

0.66

rite

0.66

BP

0.65

 Myster

0.65

é¾įå¥ĳå£«

0.65

ablishment

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.