INDEX

Explanations

words that indicate concern or issues of worry

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

New Auto-Interp

Configuration

ctigges/pythia-70m-deduped__res-sm_processed/1-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.1.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

itzer

-1.72

arest

-1.66

jour

-1.66

 honors

-1.51

 veloc

-1.50

chen

-1.43

hearted

-1.41

hoe

-1.40

FILES

-1.37

>&

-1.37

POSITIVE LOGITS

igned

1.72

ively

1.71

ted

1.70

ingly

1.66

edly

1.63

otoxin

1.59

ogenicity

1.58

ussed

1.56

iom

1.50

 areas

1.48

Activations Density 0.128%

words that indicate concern or issues of worry

No Comments

No Known Activations