INDEX

Explanations

references to violent actions resulting in death

oai_token-act-pair · gpt-3.5-turbo

references to the act of killing

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Aerospace

-0.74

rep

-0.67

Jer

-0.66

anwhile

-0.66

orthy

-0.66

Discuss

-0.65

fty

-0.64

ellation

-0.64

Altern

-0.64

MpServer

-0.63

POSITIVE LOGITS

 spree

1.28

houses

0.92

 rampage

0.88

 killings

0.85

 killing

0.83

 murdering

0.82

 deaths

0.77

 murders

0.77

house

0.76

 massac

0.76

Activations Density 0.011%

references to violent actions resulting in death

references to the act of killing

No Comments

No Known Activations