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

izontal

-0.90

Station

-0.74

GA

-0.73

Init

-0.71

DS

-0.71

Den

-0.70

GE

-0.70

ureau

-0.69

GI

-0.68

Ign

-0.68

POSITIVE LOGITS

 shortest

0.72

 violin

0.67

 happiest

0.64

 Samson

0.63

 benef

0.63

 theoret

0.62

fur

0.62

 rounds

0.61

 surpr

0.61

ly

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.