INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Beg

-0.68

Som

-0.65

umar

-0.64

DIY

-0.62

won

-0.62

 Poster

-0.61

 Mell

-0.60

 salv

-0.59

 Guant

-0.59

jee

-0.59

POSITIVE LOGITS

channelAvailability

0.87

sembly

0.80

fox

0.75

vana

0.73

blance

0.73

BOOK

0.72

ĪĴ

0.72

 ................

0.71

isSpecialOrderable

0.70

NESS

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.