INDEX

Explanations

phrases related to continuous processes or improvement

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

assadors

-1.63

owed

-1.62

 convinced

-1.58

chers

-1.54

derer

-1.50

 daddy

-1.50

 adorable

-1.41

 instructed

-1.40

 brothers

-1.39

 Chief

-1.36

POSITIVE LOGITS

£

2.10

ness

1.90

uration

1.90

eten

1.84

Ļª

1.76

©

1.75

Ł

1.73

Īĺ

1.70

idad

1.70

itÃ¤t

1.69

Activations Density 0.074%

phrases related to continuous processes or improvement

No Comments

No Known Activations