INDEX

Explanations

phrases related to controversial or divisive topics and imagery

oai_token-act-pair · gpt-3.5-turbo

elements depicting same-sex relationships or LGBTQ+ themes

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

Pwr

-0.81

 Emails

-0.76

Initial

-0.74

WAYS

-0.73

 Applications

-0.72

asons

-0.72

bors

-0.71

rences

-0.71

imester

-0.70

effective

-0.70

POSITIVE LOGITS

 nude

1.34

 grinning

1.31

 smiling

1.30

 naked

1.27

 silhou

1.27

 decap

1.26

 bearded

1.24

 silhouette

1.20

 clothed

1.17

 caricature

1.16

Activations Density 0.461%

phrases related to controversial or divisive topics and imagery

elements depicting same-sex relationships or LGBTQ+ themes

No Comments

No Known Activations