INDEX

Explanations

descriptions of places or periods where people have lived

oai_token-act-pair · gpt-3.5-turbo

references to the concept of living or residence

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

 icing

-0.75

tracking

-0.62

Add

-0.59

amin

-0.59

Add

-0.58

 announcing

-0.58

idding

-0.57

cial

-0.57

raction

-0.57

ILS

-0.57

POSITIVE LOGITS

 lived

3.47

 resided

2.09

lived

2.00

 lives

1.66

 living

1.54

 inhabited

1.51

 live

1.49

 survived

1.40

living

1.37

 died

1.37

Activations Density 0.015%

descriptions of places or periods where people have lived

references to the concept of living or residence

No Comments

No Known Activations