INDEX

Explanations

words related to bullying or coercion

oai_token-act-pair · gpt-3.5-turbo

references to intimidation or coercive tactics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Canal

-0.78

Fix

-0.67

nas

-0.66

 Lilly

-0.66

Universal

-0.65

spring

-0.64

 Doctors

-0.64

 Fault

-0.64

Ep

-0.63

 MÃ©

-0.63

POSITIVE LOGITS

 intimidation

3.46

 intimidate

2.93

 intimid

2.65

 intimidating

2.55

 intimidated

2.51

 extortion

1.75

 coercion

1.72

 blackmail

1.58

 impunity

1.51

 coer

1.48

Activations Density 0.038%

words related to bullying or coercion

references to intimidation or coercive tactics

No Comments

No Known Activations