INDEX

Explanations

statements related to controversial or scandalous behavior

oai_token-act-pair · gpt-3.5-turbo

derogatory terms and inflammatory statements aimed at individuals or groups

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 partName

-0.69

 Printed

-0.62

 anecd

-0.60

cellaneous

-0.59

icion

-0.57

 Quarterly

-0.57

 unfocusedRange

-0.57

 confir

-0.57

 aback

-0.57

 misunder

-0.56

POSITIVE LOGITS

 raping

0.88

 raped

0.77

 rape

0.76

 rapist

0.70

sex

0.70

 adultery

0.69

 vagina

0.69

 genitals

0.69

 sexually

0.68

 rapists

0.67

Activations Density 1.763%

statements related to controversial or scandalous behavior

derogatory terms and inflammatory statements aimed at individuals or groups

No Comments

No Known Activations