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

emis

-0.74

 coils

-0.70

aganda

-0.70

 caves

-0.69

 mast

-0.66

 radios

-0.66

 viol

-0.65

 dioxide

-0.65

 rods

-0.64

rint

-0.64

POSITIVE LOGITS

venge

0.84

avour

0.82

League

0.75

colour

0.74

Toronto

0.73

Manchester

0.71

prise

0.70

kj

0.69

Game

0.68

matter

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.