INDEX

Explanations

proper nouns and commercial product names

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Configuration

neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oÄŁ

-0.55

alle

-0.53

Balt

-0.52

 trave

-0.51

uth

-0.51

 concess

-0.49

allah

-0.49

ursed

-0.48

©¶æ

-0.48

ather

-0.48

POSITIVE LOGITS

 sibling

0.53

 sister

0.53

 successor

0.52

 mascot

0.52

ieth

0.51

Tik

0.51

 sponsor

0.51

 Polaris

0.51

 sequel

0.51

 Newt

0.50

Activations Density 7.035%

proper nouns and commercial product names

No Comments

No Known Activations