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

isen

-0.87

undai

-0.86

tymology

-0.84

anke

-0.84

assic

-0.81

ecided

-0.78

leep

-0.76

uden

-0.76

anmar

-0.76

avorite

-0.75

POSITIVE LOGITS

 fixed

1.21

 Fixed

0.80

fixed

0.75

FIX

0.74

ITAL

0.70

 charge

0.67

 expense

0.64

IFIED

0.63

Fixed

0.63

 Spray

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.