INDEX

Explanations

references to disciplinary actions or behaviors

oai_token-act-pair · gpt-3.5-turbo

references to rules and control mechanisms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

prop

-0.65

 Alexandria

-0.64

swick

-0.63

ods

-0.63

 Federation

-0.63

ergy

-0.62

Jane

-0.62

AIR

-0.62

Vox

-0.61

cop

-0.61

POSITIVE LOGITS

 discipline

1.42

cipline

1.11

 disciplined

1.05

 discipl

1.01

 disciplines

0.91

 Discipline

0.90

cipl

0.85

 mathemat

0.82

utic

0.82

 restraint

0.82

Activations Density 0.013%

references to disciplinary actions or behaviors

references to rules and control mechanisms

No Comments

No Known Activations