INDEX

Explanations

phrases related to potential risks or dangers

oai_token-act-pair · gpt-3.5-turbo

phrases indicating potential consequences or outcomes

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

bern

-0.89

baugh

-0.89

ablishment

-0.85

aten

-0.70

antry

-0.70

tein

-0.69

raint

-0.68

uart

-0.68

emen

-0.67

otle

-0.67

POSITIVE LOGITS

 habitable

0.99

 lifes

0.97

 jeopard

0.96

 hazardous

0.92

 lethal

0.90

 disastrous

0.88

 limitless

0.87

 carcin

0.86

 dangerous

0.85

 harmful

0.85

Activations Density 0.057%

phrases related to potential risks or dangers

phrases indicating potential consequences or outcomes

No Comments

No Known Activations