INDEX

Explanations

phrases related to terrorist activities or individuals

oai_token-act-pair · gpt-3.5-turbo

references to terrorists

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

shire

-0.70

irth

-0.69

ership

-0.67

ered

-0.67

times

-0.67

pel

-0.65

NRS

-0.65

bred

-0.65

Wheel

-0.65

laus

-0.64

POSITIVE LOGITS

 terrorists

1.07

 bombers

0.97

apons

0.91

 mastermind

0.90

 terrorist

0.88

abad

0.88

 detonated

0.85

 gunmen

0.85

terrorist

0.84

 attackers

0.83

Activations Density 0.010%

phrases related to terrorist activities or individuals

references to terrorists

No Comments

No Known Activations