INDEX

Explanations

phrases related to protection, safety, and defense

oai_token-act-pair · gpt-3.5-turbo

instances of threats or dangers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Scotland

-0.77

Enjoy

-0.76

Finish

-0.69

 Joined

-0.68

Team

-0.68

served

-0.67

thood

-0.67

 consulted

-0.66

Header

-0.64

âĦ¢:

-0.64

POSITIVE LOGITS

 onslaught

1.39

 scourge

1.37

rav

1.23

 horrors

1.22

 evils

1.20

 dangers

1.19

 worst

1.18

 effects

1.18

 brunt

1.17

 wrath

1.16

Activations Density 0.264%

phrases related to protection, safety, and defense

instances of threats or dangers

No Comments

No Known Activations