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

oho

-0.78

Georg

-0.76

gae

-0.75

EGA

-0.73

roy

-0.73

Kenn

-0.72

aimon

-0.72

alian

-0.71

rador

-0.70

ois

-0.70

POSITIVE LOGITS

 Preferred

0.77

iring

0.72

 Weight

0.69

 Digest

0.68

 Pitch

0.67

 Converted

0.66

 Feed

0.65

 Favorite

0.64

doctor

0.64

weight

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.