INDEX

Explanations

phrases related to violent actions, particularly murder

oai_token-act-pair · gpt-3.5-turbo

instances of the word "murdered" and discussions surrounding violent death

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

issue

-0.78

rium

-0.71

lag

-0.71

CON

-0.69

worthiness

-0.68

wcsstore

-0.68

yrinth

-0.67

ffic

-0.67

regulated

-0.66

annis

-0.65

POSITIVE LOGITS

 murdered

0.93

 spree

0.89

 throats

0.80

DAQ

0.75

 murders

0.75

 murdering

0.73

adoes

0.73

soever

0.72

 murder

0.72

 innoc

0.71

Activations Density 0.024%

phrases related to violent actions, particularly murder

instances of the word "murdered" and discussions surrounding violent death

No Comments

No Known Activations