INDEX

Explanations

nouns related to information manipulation and persuasion

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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Owner

-0.72

{{

-0.69

 touched

-0.68

Ö

-0.68

Oracle

-0.67

 Woods

-0.67

--+

-0.66

atta

-0.66

SUN

-0.65

eca

-0.65

POSITIVE LOGITS

 propaganda

3.42

 disinformation

2.26

aganda

2.09

 misinformation

1.74

 indoctr

1.66

 propag

1.48

 censorship

1.40

 leaflets

1.36

 ideology

1.36

 espionage

1.35

Activations Density 0.020%

nouns related to information manipulation and persuasion

references to propaganda

No Comments

No Known Activations