INDEX

Explanations

mentions of geographic locations, particularly towns and cities

oai_token-act-pair · gpt-3.5-turbo

proper nouns related to locations and geographic entities

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

 smoot

-0.75

 Reincarn

-0.69

 crank

-0.69

 positives

-0.67

 joke

-0.65

 chore

-0.63

 sque

-0.63

 thumbs

-0.63

 paranormal

-0.63

 pedals

-0.62

POSITIVE LOGITS

apan

1.03

ulia

1.02

opolis

0.99

ussia

0.98

areth

0.96

akh

0.96

onia

0.94

anya

0.93

ua

0.92

awa

0.92

Activations Density 0.302%

mentions of geographic locations, particularly towns and cities

proper nouns related to locations and geographic entities

No Comments

No Known Activations