INDEX

Explanations

phrases related to violent events, specifically shootings

oai_token-act-pair · gpt-3.5-turbo

references to incidents of gun violence or shootings

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

hw

-0.93

GY

-0.83

Label

-0.80

undai

-0.79

hern

-0.77

ateg

-0.77

redo

-0.76

ebook

-0.74

atom

-0.73

Cola

-0.71

POSITIVE LOGITS

 spree

1.31

 rampage

1.02

 shooting

0.87

 shootings

0.82

 scenes

0.81

 shoot

0.80

powder

0.79

 Shoot

0.79

nikov

0.77

 scene

0.76

Activations Density 0.022%

phrases related to violent events, specifically shootings

references to incidents of gun violence or shootings

No Comments

No Known Activations