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

 Liberty

-0.70

 Martial

-0.63

plex

-0.62

ioned

-0.62

((

-0.62

mut

-0.61

 Domin

-0.61

ellect

-0.61

 courts

-0.60

natural

-0.58

POSITIVE LOGITS

ergic

0.76

interstitial

0.75

DERR

0.73

alon

0.73

oshi

0.72

usha

0.71

Ĥİ

0.68

 Presents

0.68

owler

0.68

abbit

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.