INDEX

Explanations

words related to remarkable or extraordinary events or entities

oai_token-act-pair · gpt-3.5-turbo

terms associated with obstacles and spectacles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Madd

-0.62

 Viet

-0.59

RW

-0.58

 zones

-0.58

 reportedly

-0.57

 drives

-0.57

 shoulder

-0.56

 emphasis

-0.56

 Levin

-0.56

DOJ

-0.56

POSITIVE LOGITS

acle

4.96

acles

4.27

acular

2.55

ifice

1.39

acl

1.28

acea

1.14

acly

1.13

ator

1.08

nam

1.06

ror

1.03

Activations Density 0.013%

words related to remarkable or extraordinary events or entities

terms associated with obstacles and spectacles

No Comments

No Known Activations