INDEX

Explanations

phrases related to religious figures or worship

oai_token-act-pair · gpt-3.5-turbo

references to a divine or blessed figure

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

 mascul

-0.76

sonian

-0.71

ilater

-0.68

 parach

-0.66

 Parsons

-0.64

CBS

-0.64

hel

-0.63

 webs

-0.62

 Simpson

-0.61

enegger

-0.61

POSITIVE LOGITS

ãĢĲ

1.01

 Sakuya

0.98

Qin

0.93

âĶĢâĶĢ

0.93

 Yuan

0.90

ãĢİ

0.88

 Qian

0.87

Luo

0.87

 Continent

0.86

Xu

0.85

Activations Density 0.759%

phrases related to religious figures or worship

references to a divine or blessed figure

No Comments

No Known Activations