INDEX

Explanations

words or punctuation that indicate a list or series

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĻĤ

-2.20

anguage

-1.98

à±į

-1.88

FromString

-1.80

wagen

-1.71

opan

-1.71

__,

-1.71

oxacin

-1.70

-1.69

rowave

-1.68

POSITIVE LOGITS

 luck

1.76

``

1.72

too

1.66

 confounding

1.56

 albeit

1.56

 thereby

1.51

which

1.50

1.49

 heck

1.49

 waiting

1.49

Activations Density 1.206%

words or punctuation that indicate a list or series

No Comments

No Known Activations