INDEX

Explanations

phrases related to principles or fundamental beliefs

oai_token-act-pair · gpt-3.5-turbo

references to principles and laws

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

minster

-0.78

otte

-0.73

essions

-0.71

IED

-0.68

ilation

-0.67

oÄŁ

-0.65

 Carbuncle

-0.64

hiba

-0.63

NetMessage

-0.63

ummer

-0.63

POSITIVE LOGITS

 principles

1.01

 underpin

0.93

 underlying

0.91

 guiding

0.90

 principle

0.89

ually

0.86

ciples

0.86

 Principles

0.82

 principals

0.79

cipled

0.78

Activations Density 0.040%

phrases related to principles or fundamental beliefs

references to principles and laws

No Comments

No Known Activations