INDEX

Explanations

words related to political propaganda

oai_token-act-pair · gpt-3.5-turbo

references to propaganda

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

kens

-0.83

hens

-0.74

kward

-0.71

nings

-0.70

Moj

-0.68

bang

-0.67

ðĿ

-0.67

ced

-0.66

Numbers

-0.66

desc

-0.66

POSITIVE LOGITS

 campaigns

1.13

 leaflets

1.05

 dissemin

1.05

aganda

1.04

 posters

0.98

 disinformation

0.95

 propaganda

0.94

 pamph

0.94

 propag

0.93

 mills

0.92

Activations Density 0.078%

words related to political propaganda

references to propaganda

No Comments

No Known Activations