INDEX

Explanations

words related to violence and killing

oai_token-act-pair · gpt-3.5-turbo

references to the concept of killing

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

BuyableInstoreAndOnline

-0.74

ORY

-0.72

Cola

-0.72

Scot

-0.72

 Depot

-0.68

 Failure

-0.68

oday

-0.68

 Collider

-0.65

DragonMagazine

-0.63

 Municip

-0.63

POSITIVE LOGITS

joy

1.05

mails

0.95

 spree

0.94

switch

0.91

blow

0.80

mong

0.75

 civilians

0.74

off

0.73

lords

0.72

fish

0.72

Activations Density 0.067%

words related to violence and killing

references to the concept of killing

No Comments

No Known Activations