INDEX

Explanations

phrases indicating legal definitions or statuses

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

New Auto-Interp

Configuration

ctigges/pythia-70m-deduped__mlp-sm_processed/0-mlp-sm

Prompts (Dashboard)

32,768 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

32,768

Data Type

torch.float32

Hook Name

blocks.0.hook_mlp_out

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

±

-2.91

ĥ½

-2.87

½

-2.83

ģ

-2.64

¸

-2.62

ĭ

-2.52

į

-2.50

®

-2.47

ĩ

-2.44

Ĳ

-2.41

POSITIVE LOGITS

 Justice

1.47

 Privacy

1.40

ulsion

1.38

ibus

1.36

orect

1.36

fraction

1.33

 wrong

1.30

olecular

1.30

 interference

1.29

 pardon

1.28

Activations Density 0.117%

phrases indicating legal definitions or statuses

No Comments

No Known Activations