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

nesota

-0.94

wagen

-0.86

conservancy

-0.76

notations

-0.72

DragonMagazine

-0.72

irus

-0.71

 Liberties

-0.71

ivot

-0.71

iverpool

-0.70

notation

-0.68

POSITIVE LOGITS

ublic

0.71

 controlled

0.66

Lou

0.64

ADS

0.63

 PRESIDENT

0.62

Ã¡

0.62

ched

0.61

 sitting

0.60

 Mondays

0.60

 Valent

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.