INDEX

Explanations

phrases related to professional roles or titles

oai_token-act-pair · gpt-3.5-turbo

job titles and roles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Ø§ÙĦ

-0.70

omething

-0.62

 diabetic

-0.61

 Toro

-0.59

 Invalid

-0.59

STDOUT

-0.59

Torrent

-0.58

 Unch

-0.57

 Delicious

-0.56

ERO

-0.54

POSITIVE LOGITS

ial

1.45

ially

1.31

ials

1.28

alties

1.25

ship

1.22

hood

1.22

SHIP

1.18

rics

1.13

cies

1.10

dom

1.10

Activations Density 0.285%

phrases related to professional roles or titles

job titles and roles

No Comments

No Known Activations