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

 Issues

-0.72

CLASSIFIED

-0.71

SEC

-0.67

 Occupations

-0.64

Vaj

-0.62

 cared

-0.62

Zub

-0.61

SourceFile

-0.61

 Cous

-0.61

 Damien

-0.60

POSITIVE LOGITS

odon

0.75

grain

0.67

berman

0.67

ricular

0.67

itionally

0.64

 bait

0.64

nesday

0.63

idon

0.63

unte

0.62

itone

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.