INDEX

Explanations

phrases related to safety advice and warnings

oai_token-act-pair · gpt-3.5-turbo

warnings and advice related to safety

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 fame

-0.69

 descendants

-0.69

 creator

-0.67

 magically

-0.66

 victorious

-0.66

 unveiling

-0.66

created

-0.64

 transformed

-0.64

 Divinity

-0.63

 Noble

-0.63

POSITIVE LOGITS

 caution

1.59

 beware

1.58

eware

1.30

 precautions

1.25

Avoid

1.23

 precaution

1.22

avoid

1.17

 vigilance

1.16

 advis

1.16

 advisable

1.16

Activations Density 1.291%

phrases related to safety advice and warnings

warnings and advice related to safety

No Comments

No Known Activations