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

 Wand

-0.73

 Instruction

-0.72

 Velvet

-0.69

Elf

-0.66

 Aerial

-0.65

 Mana

-0.65

 Elves

-0.65

 veiled

-0.64

 Abyssal

-0.64

 Wilkinson

-0.63

POSITIVE LOGITS

aunder

0.79

culosis

0.79

 mosqu

0.75

 WATCHED

0.74

mson

0.72

enter

0.71

uckland

0.71

dinand

0.71

uncture

0.70

ricanes

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.