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

td

-0.72

division

-0.71

Dispatch

-0.71

ungle

-0.70

odon

-0.69

rador

-0.66

 columns

-0.66

Nusra

-0.63

column

-0.63

ruary

-0.62

POSITIVE LOGITS

natureconservancy

0.95

 haircut

0.77

 EntityItem

0.76

 environmentally

0.73

idge

0.70

 ILCS

0.69

bilt

0.67

kees

0.66

rust

0.65

 handshake

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.