INDEX

Explanations

phrases related to hate speech and hate crimes

oai_token-act-pair · gpt-3.5-turbo

references to hate and hate speech

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

ufact

-0.82

æ©Ł

-0.82

UNCH

-0.80

å§«

-0.79

aver

-0.76

idges

-0.73

Decre

-0.71

é¾įå

-0.71

 Tablet

-0.70

ODE

-0.70

POSITIVE LOGITS

fulness

1.15

fully

1.12

 crimes

0.96

ful

0.91

 vengeance

0.86

 hate

0.86

hate

0.79

 retaliation

0.78

ãĥĨ

0.78

bre

0.78

Activations Density 0.017%

phrases related to hate speech and hate crimes

references to hate and hate speech

No Comments

No Known Activations