INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

goo

-0.83

rity

-0.69

urg

-0.68

burg

-0.65

urt

-0.65

 orgasm

-0.64

 promotion

-0.63

haar

-0.62

icide

-0.62

 flowing

-0.62

POSITIVE LOGITS

é¾įå

0.91

 Cosponsors

0.89

ãĤ¢ãĥ«

0.87

æķ

0.85

ãĥĩãĤ£

0.83

xtap

0.80

 Phar

0.78

éģ

0.77

æĥ

0.76

ython

0.74

Activations Density 0.000%

No Known Activations

This feature has no known activations.