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

uve

-0.70

vre

-0.69

otive

-0.68

eatured

-0.66

BuyableInstoreAndOnline

-0.64

cad

-0.63

EngineDebug

-0.63

»Ĵ

-0.62

chal

-0.61

cellaneous

-0.61

POSITIVE LOGITS

adelphia

0.76

 libertarians

0.72

ombat

0.71

 economist

0.64

eton

0.64

 Scotia

0.63

Rao

0.62

cro

0.62

ONES

0.61

 Parties

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.