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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ope

-0.85

aina

-0.75

ayed

-0.72

ul

-0.71

arah

-0.70

Cla

-0.70

 Lite

-0.68

riet

-0.67

opes

-0.66

hel

-0.65

POSITIVE LOGITS

nces

0.88

lihood

0.87

 magnet

0.76

 conclud

0.73

xual

0.70

 traject

0.70

Leaks

0.68

 disse

0.67

 CONFIG

0.66

 menacing

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.