INDEX

Explanations

references to violent actions resulting in harm or death

oai_token-act-pair · gpt-3.5-turbo

references to violent deaths and fatalities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

wcsstore

-0.90

channelAvailability

-0.82

hist

-0.77

DragonMagazine

-0.76

soType

-0.74

arist

-0.72

 deleg

-0.70

DIR

-0.70

Han

-0.67

hatt

-0.66

POSITIVE LOGITS

 spree

0.94

 rampage

0.80

 gunshot

0.78

 Sandra

0.76

 stabbing

0.75

nsics

0.74

llah

0.73

 psychiat

0.72

 manslaughter

0.72

 Trayvon

0.71

Activations Density 0.083%

references to violent actions resulting in harm or death

references to violent deaths and fatalities

No Comments

No Known Activations