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

wcs

-0.78

recomm

-0.68

Water

-0.64

utor

-0.63

Reviewed

-0.63

 water

-0.63

waters

-0.63

legates

-0.63

 Reserv

-0.61

ivers

-0.61

POSITIVE LOGITS

1.00

 Thou

0.76

Â·Â·

0.72

 Elon

0.70

SPONSORED

0.68

 Koen

0.67

Ely

0.66

 Cain

0.66

 Idle

0.64

 Bundy

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.