INDEX

Explanations

phrases related to authority or official positions

oai_token-act-pair · gpt-3.5-turbo

the word "presume" and its variations, indicating uncertainty or assumption

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ISM

-0.66

 Abyssal

-0.66

 Welsh

-0.63

 Squid

-0.63

 Lions

-0.62

 Italians

-0.61

bean

-0.61

 dots

-0.61

 Bruins

-0.60

 IMAGES

-0.60

POSITIVE LOGITS

chool

1.36

cription

1.34

ervation

1.30

umed

1.25

byter

1.24

erver

1.18

erving

1.17

idents

1.16

erves

1.15

umably

1.15

Activations Density 0.037%

phrases related to authority or official positions

the word "presume" and its variations, indicating uncertainty or assumption

No Comments

No Known Activations