INDEX

Explanations

news-related keywords and phrases, specifically focusing on video content recommendations

oai_token-act-pair · gpt-3.5-turbo

phrases indicating video replay or viewing

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

doms

-0.67

Ļ½

-0.66

pora

-0.65

kers

-0.63

zbek

-0.63

 subsistence

-0.62

 plat

-0.62

enium

-0.62

ker

-0.61

krit

-0.60

POSITIVE LOGITS

 Replay

0.98

 WATCHED

0.89

 Videos

0.77

 WATCH

0.73

 Thumbnails

0.72

lements

0.71

iece

0.70

INGS

0.68

acher

0.68

ique

0.66

Activations Density 0.014%

news-related keywords and phrases, specifically focusing on video content recommendations

phrases indicating video replay or viewing

No Comments

No Known Activations