INDEX

Explanations

common phrases related to favoriting, rating, and sharing online content

oai_token-act-pair · gpt-3.5-turbo

mentions of favorites or preferred items

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

aea

-0.71

utical

-0.69

raud

-0.67

 Frie

-0.66

aucas

-0.65

 Centauri

-0.65

kaya

-0.65

lessness

-0.64

urg

-0.63

ufact

-0.63

POSITIVE LOGITS

 favorite

1.56

favorite

1.17

 Favorite

1.17

 favourite

1.06

Favorite

0.96

 favorites

0.95

¥µ

0.80

 Favor

0.70

Reviewer

0.69

 favored

0.68

Activations Density 0.016%

common phrases related to favoriting, rating, and sharing online content

mentions of favorites or preferred items

No Comments

No Known Activations