INDEX

Explanations

words related to crimes, especially homicides

oai_token-act-pair · gpt-3.5-turbo

references to homicide

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

ais

-0.82

Cola

-0.76

amina

-0.75

kef

-0.73

atche

-0.71

hatt

-0.70

enna

-0.67

ube

-0.67

itsch

-0.67

endon

-0.67

POSITIVE LOGITS

 homicide

1.14

 homicides

0.96

 spree

0.83

icide

0.75

 victim

0.75

 prevention

0.74

 deaths

0.74

 victims

0.73

 Deaths

0.70

icides

0.70

Activations Density 0.005%

words related to crimes, especially homicides

references to homicide

No Comments

No Known Activations