INDEX

Explanations

references to spices, specifically when related to illegal substances or food ingredients

oai_token-act-pair · gpt-3.5-turbo

references to various types of spice

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inel

-0.81

aft

-0.72

 Bundes

-0.72

Gou

-0.71

OU

-0.67

 Miliband

-0.67

 ..........

-0.67

 Hannah

-0.66

 MÃ©

-0.66

 Zack

-0.65

POSITIVE LOGITS

 spice

3.92

 Spice

2.74

 spices

2.52

 curry

1.94

 spicy

1.66

 cinnamon

1.61

 seasoning

1.56

 perfume

1.15

 chili

1.14

 cloves

1.14

Activations Density 0.033%

references to spices, specifically when related to illegal substances or food ingredients

references to various types of spice

No Comments

No Known Activations