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

icle

-0.77

orer

-0.72

obe

-0.68

uel

-0.68

piration

-0.67

yrim

-0.65

ree

-0.65

ading

-0.65

sey

-0.64

ghai

-0.64

POSITIVE LOGITS

 elig

0.67

 eagle

0.67

 Ezek

0.67

Extra

0.66

Merit

0.64

 PACs

0.64

 Hutch

0.64

 EDITION

0.64

 issu

0.64

è»

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.