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

ĸļ

-1.07

ado

-0.82

tnc

-0.75

icer

-0.75

¿½

-0.75

attach

-0.74

issance

-0.72

hart

-0.69

cial

-0.68

 Inquiry

-0.68

POSITIVE LOGITS

 Roof

0.73

 Lovecraft

0.69

KK

0.69

KNOWN

0.67

 gamb

0.65

opol

0.65

 Lump

0.65

 Twisted

0.64

 twisted

0.64

Kus

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.