INDEX

Explanations

the word "giants" when discussing influential entities or powerful organizations

oai_token-act-pair · gpt-3.5-turbo

references to large, influential companies or organizations

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

 Assignment

-0.77

tenance

-0.66

ader

-0.65

âĸĪâĸĪ

-0.64

 assigned

-0.63

 volunteer

-0.63

oter

-0.62

 Degree

-0.61

 assignment

-0.60

 voter

-0.60

POSITIVE LOGITS

 giants

3.85

 pioneers

1.44

 legends

1.43

 Giants

1.37

 giant

1.33

 champions

1.29

 dwarves

1.28

 gods

1.24

 goblins

1.23

 dinosaurs

1.22

Activations Density 0.021%

the word "giants" when discussing influential entities or powerful organizations

references to large, influential companies or organizations

No Comments

No Known Activations