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

âĢ¢âĢ¢

-0.85

¥µ

-0.71

 Rudolph

-0.66

 Kramer

-0.63

 Younger

-0.62

 Berks

-0.61

ĵĺ

-0.60

 Buddy

-0.59

 BART

-0.58

Wow

-0.58

POSITIVE LOGITS

aldo

0.83

folk

0.80

nia

0.70

 redes

0.68

ny

0.67

utherland

0.66

iege

0.65

 isEnabled

0.64

ieri

0.64

 tabl

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.