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

her

-0.69

oglu

-0.63

 messenger

-0.62

SEE

-0.61

 SERV

-0.61

 geop

-0.60

accompan

-0.60

Footnote

-0.60

Clean

-0.59

 reservations

-0.59

POSITIVE LOGITS

 Nanto

0.78

apolis

0.71

lled

0.70

itialized

0.68

00000000

0.65

pha

0.65

ãĤ±

0.63

 Aether

0.63

¥µ

0.63

 Norn

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.