INDEX

Explanations

terms related to dangerous situations or events

oai_token-act-pair · gpt-3.5-turbo

terms related to lethal incidents or conditions

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

 Simpl

-0.74

Bid

-0.65

RFC

-0.64

Uni

-0.64

manager

-0.63

 Lerner

-0.62

lite

-0.62

ijn

-0.62

QC

-0.62

uci

-0.62

POSITIVE LOGITS

 deadly

3.55

 lethal

2.45

 Deadly

2.22

 deadliest

2.19

 fatal

2.07

 dangerous

1.95

 murderous

1.76

 devastating

1.66

 poisonous

1.64

 hazardous

1.49

Activations Density 0.013%

terms related to dangerous situations or events

terms related to lethal incidents or conditions

No Comments

No Known Activations