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

SourceFile

-0.88

phthal

-0.79

Nit

-0.75

universal

-0.74

hered

-0.71

perties

-0.69

health

-0.69

gently

-0.68

resistant

-0.67

ITY

-0.65

POSITIVE LOGITS

 Moonlight

0.81

 Poles

0.77

 Caldwell

0.75

 Powder

0.66

 Appalachian

0.66

 Parade

0.65

 Miami

0.63

CW

0.62

scape

0.62

 Balk

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.