INDEX

Explanations

names of countries

oai_token-act-pair · gpt-3.5-turbo

proper nouns, specifically names of movies and geographical locations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

idated

-0.41

ividual

-0.38

idate

-0.35

enegger

-0.34

allery

-0.34

 Daytona

-0.34

innon

-0.33

uilt

-0.33

 Honour

-0.33

Jol

-0.32

POSITIVE LOGITS

etheless

0.59

NetMessage

0.51

terness

0.45

ĵĺ

0.45

natureconservancy

0.44

minist

0.36

 brill

0.35

 hous

0.34

 suscept

0.34

-->

0.34

Activations Density 12.739%

names of countries

proper nouns, specifically names of movies and geographical locations

No Comments

No Known Activations