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

esta

-0.99

apo

-0.97

-0.91

enta

-0.90

ailing

-0.79

quick

-0.78

kay

-0.78

ugi

-0.77

aps

-0.75

agra

-0.73

POSITIVE LOGITS

 Scientology

0.79

 Scient

0.77

 Dragonbound

0.74

 adoptive

0.69

 Dresden

0.69

å°Ĩ

0.68

 Witcher

0.68

 molecular

0.68

 contag

0.67

 Huss

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.