INDEX

Explanations

violence-related phrases, especially involving physical altercations

oai_token-act-pair · gpt-3.5-turbo

events related to violent confrontations or altercations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Reincarn

-0.89

ongyang

-0.80

 pilgrimage

-0.79

ships

-0.79

obyl

-0.76

ship

-0.76

 advent

-0.75

soDeliveryDate

-0.75

 hiber

-0.73

 Inher

-0.72

POSITIVE LOGITS

 bystanders

1.39

 onlook

1.14

 paramedics

1.12

 witnesses

1.11

 gunshots

1.09

 SWAT

1.09

 officers

1.06

 deputies

1.04

 cops

1.03

 assailants

1.01

Activations Density 0.375%

violence-related phrases, especially involving physical altercations

events related to violent confrontations or altercations

No Comments

No Known Activations