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

 sidx

-0.77

Mek

-0.76

meg

-0.70

 Dele

-0.68

 pornographic

-0.63

Kik

-0.62

 backdoor

-0.61

Vaj

-0.61

 sqor

-0.61

 monet

-0.61

POSITIVE LOGITS

uez

0.77

tymology

0.72

Owner

0.70

edIn

0.70

uracy

0.69

onder

0.68

qv

0.68

Availability

0.68

Condition

0.67

#$

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.