INDEX

Explanations

references to locations or nations, particularly Germany

oai_token-act-pair · gpt-3.5-turbo

mentions of Germany in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Vaugh

-0.74

aqu

-0.72

idav

-0.71

inventoryQuantity

-0.68

ascript

-0.65

aun

-0.63

 snake

-0.62

fill

-0.61

 Scorpion

-0.60

 subscript

-0.60

POSITIVE LOGITS

 Germany

3.47

Germany

3.22

 Germans

2.31

 German

2.30

German

2.28

 Austria

2.22

 Poland

2.18

 Belgium

2.17

 France

2.16

 Berlin

2.15

Activations Density 0.016%

references to locations or nations, particularly Germany

mentions of Germany in various contexts

No Comments

No Known Activations