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

 inspires

-0.82

aer

-0.82

 acceler

-0.71

 awaits

-0.68

azel

-0.68

 executes

-0.67

 rehearsal

-0.65

 filming

-0.65

 invests

-0.64

 unfolds

-0.63

POSITIVE LOGITS

 Cheong

0.74

ignty

0.73

 Slot

0.73

galitarian

0.70

 Argon

0.68

 Centauri

0.68

 Scotia

0.67

alan

0.66

uphem

0.66

angan

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.