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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Leopard

-0.76

ISION

-0.68

 Lions

-0.67

 Barg

-0.66

 Jaguars

-0.65

 Nadu

-0.65

addy

-0.64

 Savings

-0.64

Sun

-0.63

 Cavaliers

-0.63

POSITIVE LOGITS

merce

0.92

illin

0.84

vere

0.84

ster

0.83

ulz

0.77

yz

0.76

acid

0.75

nder

0.73

oppable

0.73

lde

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.