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

proof

-0.68

ords

-0.65

rored

-0.64

ais

-0.64

 Skydragon

-0.63

fulness

-0.63

imony

-0.63

 Papers

-0.62

lessness

-0.62

fare

-0.61

POSITIVE LOGITS

ansky

0.78

NetMessage

0.77

arnaev

0.76

Thumbnail

0.73

ä¹

0.68

aukee

0.66

culosis

0.65

ãĤ¨ãĥ«

0.64

£ı

0.63

anski

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.