INDEX

Explanations

references to intense emotional states or behavior, particularly related to the concept of madness

oai_token-act-pair · gpt-3.5-turbo

instances of the word "mad" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Downloadha

-0.74

ãģ¯

-0.69

akeru

-0.64

_>

-0.63

VEL

-0.62

Lay

-0.62

 polled

-0.60

AFB

-0.60

Ĥİ

-0.59

advertisement

-0.59

POSITIVE LOGITS

cap

1.26

agascar

1.11

der

1.11

rid

1.06

ness

1.06

ly

1.00

man

0.96

men

0.95

house

0.95

cow

0.95

Activations Density 0.018%

references to intense emotional states or behavior, particularly related to the concept of madness

instances of the word "mad" in various contexts

No Comments

No Known Activations