INDEX

Explanations

references to intelligence or being clever

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

wegian

-1.73

 Suite

-1.49

 Marathon

-1.41

ktop

-1.35

jas

-1.33

 muse

-1.33

clusive

-1.33

ede

-1.31

 darker

-1.30

 href

-1.30

POSITIVE LOGITS

ĻĤ

2.38

 contracts

1.96

ģ

1.90

Ħ

1.81

Ģ

1.76

itÃ©s

1.60

Īĺ

1.59

ĥ½

1.58

İ

1.56

ĸ

1.56

Activations Density 0.105%

references to intelligence or being clever

No Comments

No Known Activations