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

onel

-0.76

rists

-0.76

ourney

-0.67

nesota

-0.67

chance

-0.64

 helic

-0.64

amera

-0.62

lder

-0.62

kas

-0.62

omal

-0.61

POSITIVE LOGITS

 WARN

0.73

 "$:/

0.70

(){

0.68

}}}

0.64

 RELE

0.64

OLD

0.63

ksh

0.62

Kot

0.62

ISO

0.62

cakes

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.