INDEX

Explanations

references to various types of vessels or ships

oai_token-act-pair · gpt-3.5-turbo

references to maritime vessels

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Lamar

-0.69

cho

-0.69

eworld

-0.69

nesota

-0.67

âĶģ

-0.66

hee

-0.64

enhagen

-0.64

andro

-0.64

 fertile

-0.64

arching

-0.63

POSITIVE LOGITS

 vessel

1.37

 vessels

1.24

 Vessel

1.15

HMS

0.93

wright

0.87

 Titanic

0.87

 Yamato

0.85

 boats

0.85

 boat

0.83

USS

0.83

Activations Density 0.008%

references to various types of vessels or ships

references to maritime vessels

No Comments

No Known Activations