INDEX

Explanations

data related to computer programming, such as code snippets and technical instructions

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Configuration

neuronpedia/gpt2-small__res_scefr-ajt/6-res_scefr-ajt

Prompts (Dashboard)

12,288 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

46,080

Data Type

torch.float32

Hook Point

blocks.6.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

apollo-research/Skylion007-openwebtext-tokenizer-gpt2

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

•Layer 6 UMAP region: e2e-recon outlier bottom right - e2e-recon

No Comments

Negative Logits

ulously

-0.50

 nomine

-0.49

uling

-0.48

 Guinea

-0.47

aturdays

-0.45

 Sawyer

-0.45

oulos

-0.44

uler

-0.42

 obligatory

-0.42

Ħ¢

-0.42

POSITIVE LOGITS

 Crimson

0.57

Trivia

0.52

cakes

0.51

 Kinnikuman

0.50

natureconservancy

0.49

 Fang

0.49

Spread

0.49

PokÃ©mon

0.47

tackle

0.47

 unbeat

0.46

Activations Density 3.291%

data related to computer programming, such as code snippets and technical instructions

No Comments

No Known Activations