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

atche

-0.88

pecially

-0.73

entials

-0.69

Fax

-0.68

iterranean

-0.65

enza

-0.64

iciency

-0.62

ulf

-0.62

inav

-0.61

ropolitan

-0.60

POSITIVE LOGITS

pot

1.99

 pots

1.26

 bowls

0.75

arijuana

0.74

 canoe

0.73

 psychedel

0.71

bnb

0.71

 Marijuana

0.70

 marijuana

0.70

bon

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.