INDEX

Explanations

proper nouns such as names of people and organizations

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 licences

-0.95

 licence

-0.92

 recognised

-0.83

 ageing

-0.80

 civilisation

-0.78

 mould

-0.78

 realised

-0.77

 endeavour

-0.77

 practise

-0.77

 envis

-0.75

POSITIVE LOGITS

 Politico

1.56

CNN

1.48

 POLITICO

1.44

 MSNBC

1.42

CNN

1.36

GOP

1.29

 Hannity

1.29

 Huckabee

1.29

 Breitbart

1.27

TRUMP

1.24

Activations Density 0.877%

proper nouns such as names of people and organizations

No Comments

No Known Activations