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

"]=>

-0.78

 Nasa

-0.75

 Mehran

-0.75

"],"

-0.70

DragonMagazine

-0.69

elf

-0.69

urg

-0.68

erva

-0.66

 Manor

-0.66

bitious

-0.66

POSITIVE LOGITS

SPONSORED

0.71

 detriment

0.68

lly

0.67

 exception

0.65

lege

0.65

paw

0.65

 lobe

0.65

bud

0.63

theless

0.62

hey

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.