INDEX

Explanations

technical terms and programming-related expressions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

ctigges/pythia-70m-deduped__res-sm_processed/2-res-sm

Prompts (Dashboard)

32,768 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

32,768

Data Type

torch.float32

Hook Name

blocks.2.hook_resid_post

Hook Layer

Architecture

standard

Context Size

128

Dataset

EleutherAI/the_pile_deduplicated

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

azine

-1.43

ÃŁe

-1.39

kerchief

-1.37

Ã©rie

-1.34

uchi

-1.34

Â®

-1.32

Ã¶s

-1.28

Â

-1.27

 alkaline

-1.23

mann

-1.20

POSITIVE LOGITS

eligible

1.57

 contests

1.55

oxford

1.52

insula

1.51

 consec

1.48

yards

1.48

dwell

1.44

/#

1.42

 Papers

1.41

»¿

1.41

Activations Density 5.342%

technical terms and programming-related expressions

No Comments

No Known Activations