INDEX

Explanations

references to taking a closer or more detailed look at something

oai_token-act-pair · gpt-3.5-turbo

phrases that prompt the reader to take action or look closely at something

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

Develop

-0.74

stories

-0.71

ambo

-0.67

 disputes

-0.64

tions

-0.64

 defects

-0.63

Ott

-0.62

enf

-0.61

 agreements

-0.61

âĢ¢âĢ¢âĢ¢âĢ¢

-0.61

POSITIVE LOGITS

 look

1.25

 peek

1.23

 stroll

1.16

sip

1.01

 glimpse

1.00

 breat

0.99

 plunge

0.96

 moment

0.96

 glance

0.96

 LOOK

0.95

Activations Density 0.055%

references to taking a closer or more detailed look at something

phrases that prompt the reader to take action or look closely at something

No Comments

No Known Activations

references to taking a closer or more detailed look at something

phrases that prompt the reader to take action or look closely at something

No Comments

No Known Activations