INDEX

Explanations

phrases or words related to violent actions, especially killing

oai_token-act-pair · gpt-3.5-turbo

references to killings and related violent acts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Cola

-0.72

isse

-0.69

rypt

-0.68

gow

-0.68

fty

-0.67

TAG

-0.67

MpServer

-0.67

Discuss

-0.66

BuyableInstoreAndOnline

-0.66

Gi

-0.65

POSITIVE LOGITS

 spree

1.23

houses

0.93

 killings

0.83

 murdering

0.80

 massac

0.79

 rampage

0.78

killing

0.77

knife

0.76

 killing

0.76

 murders

0.76

Activations Density 0.023%

phrases or words related to violent actions, especially killing

references to killings and related violent acts

No Comments

No Known Activations