INDEX

Explanations

words related to power dynamics, influence, and manipulation

oai_token-act-pair · gpt-3.5-turbo

references to social and political manipulation

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

ridor

-0.81

Wave

-0.74

_{

-0.72

ato

-0.69

beam

-0.68

atown

-0.67

 awoken

-0.66

auntlet

-0.66

adobe

-0.64

 Stephenson

-0.62

POSITIVE LOGITS

 accomplishments

1.42

 failings

1.37

 evils

1.31

 plight

1.27

 shortcomings

1.26

 injust

1.23

 virtues

1.21

 atrocities

1.21

 wrongdoing

1.20

 achievements

1.20

Activations Density 0.315%

words related to power dynamics, influence, and manipulation

references to social and political manipulation

No Comments

No Known Activations