INDEX

Explanations

highly descriptive terms related to politics and individuals involved in political activities

oai_token-act-pair · gpt-3.5-turbo

terms related to divisive political figures

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

 Simulator

-0.83

eria

-0.74

é¾įåĸļå£«

-0.73

crow

-0.73

ercise

-0.71

ruction

-0.71

atown

-0.70

mage

-0.70

ossession

-0.69

anguage

-0.69

POSITIVE LOGITS

 presidents

1.39

 personalities

1.39

 secretaries

1.38

 politicians

1.36

 ambassadors

1.34

 nominees

1.33

 leaders

1.32

 advisers

1.31

 figures

1.31

 celebrities

1.30

Activations Density 0.359%

highly descriptive terms related to politics and individuals involved in political activities

terms related to divisive political figures

No Comments

No Known Activations