INDEX

Explanations

phrases related to violent incidents, specifically shootings and gunmen

oai_token-act-pair · gpt-3.5-turbo

references to shooting incidents or violent events

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

GY

-0.76

undai

-0.74

adian

-0.70

eday

-0.69

rian

-0.68

amel

-0.67

hw

-0.67

Label

-0.67

OLOGY

-0.65

anium

-0.65

POSITIVE LOGITS

 spree

1.20

 rampage

1.10

powder

0.89

nikov

0.88

 shootings

0.85

 deaths

0.84

 massacre

0.81

 Shooter

0.81

 shooter

0.81

 shooting

0.81

Activations Density 0.039%

phrases related to violent incidents, specifically shootings and gunmen

references to shooting incidents or violent events

No Comments

No Known Activations