INDEX

Explanations

mentions of religious figures, specifically Popes

oai_token-act-pair · gpt-3.5-turbo

mentions of Popes and the Vatican

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ript

-0.81

yrinth

-0.78

é¾

-0.74

schild

-0.73

skirts

-0.72

mble

-0.70

nesota

-0.68

rha

-0.68

nel

-0.67

hooting

-0.67

POSITIVE LOGITS

 Francis

1.29

 Benedict

1.02

esses

0.84

 Clement

0.83

 infall

0.81

Franc

0.79

 Pope

0.79

 pope

0.79

otle

0.78

Leo

0.74

Activations Density 0.024%

mentions of religious figures, specifically Popes

mentions of Popes and the Vatican

No Comments

No Known Activations