INDEX

Explanations

phrases indicating to keep an eye on something or someone

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate keeping watch or paying attention

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

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

minus

-0.68

theless

-0.67

 congr

-0.66

 Violent

-0.64

 Vide

-0.64

stood

-0.63

channelAvailability

-0.63

emonium

-0.62

 Resistance

-0.60

ities

-0.59

POSITIVE LOGITS

 sockets

0.99

ight

0.89

bre

0.86

eye

0.84

eyed

0.83

ovember

0.79

 noses

0.79

 nose

0.78

bridge

0.78

Eye

0.78

Activations Density 0.016%

phrases indicating to keep an eye on something or someone

phrases that indicate keeping watch or paying attention

No Comments

No Known Activations