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

 Dino

-0.69

azo

-0.67

Mb

-0.66

Eva

-0.62

eva

-0.62

xf

-0.61

 Niet

-0.61

âĪĴ

-0.60

 Moons

-0.60

 narrator

-0.59

POSITIVE LOGITS

hang

0.75

ajor

0.73

omore

0.69

 arsen

0.63

Cod

0.62

SHIP

0.62

tis

0.62

agy

0.62

AUD

0.62

skirts

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.