INDEX

Explanations

numerical or formatting indicators in structured texts

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits



-0.51

ManagedBean

-0.49

rrggbb

-0.48

TemporalType

-0.47

plata

-0.46

AsUp

-0.46

 intptr

-0.46

 gynhyrchwyd

-0.45

 ")");

-0.44

haway

-0.44

POSITIVE LOGITS

 Khusus

0.41

 Tamaño

0.39

Referències

0.39

entibus

0.39

 fondament

0.38

nictví

0.37

 Lingkungan

0.37

 Komunikasi

0.37

 Ukuran

0.36

 Biaya

0.36

Activations Density 0.000%

numerical or formatting indicators in structured texts

No Comments

No Known Activations