INDEX

Explanations

phrases indicating refusal or resistance

oai_token-act-pair · gpt-3.5-turbo

instances of the word "refused" and related phrases indicating non-compliance or resistance

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

issance

-0.86

////////////////////////////////

-0.80

renheit

-0.66

sav

-0.64

Ca

-0.64

calling

-0.64

 Thumbnails

-0.63

imal

-0.63

————————

-0.63

 Canary

-0.63

POSITIVE LOGITS

 acknowledge

1.38

bud

1.34

 concede

1.30

 cooperate

1.28

 acknow

1.23

 accept

1.21

 participate

1.18

 comply

1.16

 obey

1.16

 admit

1.15

Activations Density 0.075%

phrases indicating refusal or resistance

instances of the word "refused" and related phrases indicating non-compliance or resistance

No Comments

No Known Activations