INDEX

Explanations

phrases related to espionage and spy activities

oai_token-act-pair · gpt-3.5-turbo

elements of espionage and intelligence activities

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

ukemia

-0.88

 NASCAR

-0.81

galitarian

-0.80

 bicycl

-0.80

 Towns

-0.78

 zoning

-0.74

bernatorial

-0.73

orest

-0.72

health

-0.72

 crowdfunding

-0.71

POSITIVE LOGITS

 spies

1.99

 espionage

1.90

KGB

1.85

spy

1.84

 spying

1.77

CIA

1.68

 informants

1.65

 intelligence

1.63

CIA

1.61

 covert

1.61

Activations Density 0.875%

phrases related to espionage and spy activities

elements of espionage and intelligence activities

No Comments

No Known Activations