INDEX

Explanations

phrases indicating something as intriguing or thought-provoking

oai_token-act-pair · gpt-3.5-turbo

descriptions of topics or content deemed intriguing or noteworthy

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

âĢ¢

-0.75

dom

-0.73

rush

-0.68

stretched

-0.67

 substitute

-0.66

ulk

-0.64

 ÂŃ

-0.63

 Pledge

-0.63

FC

-0.62

estic

-0.62

POSITIVE LOGITS

 interesting

3.27

interesting

2.56

 fascinating

2.31

 intriguing

2.30

 Interesting

2.19

 exciting

1.86

 amusing

1.84

 interestingly

1.60

 noteworthy

1.60

 important

1.53

Activations Density 0.028%

phrases indicating something as intriguing or thought-provoking

descriptions of topics or content deemed intriguing or noteworthy

No Comments

No Known Activations