INDEX

Explanations

Japanese particles and conjunctions indicating grammatical relationships

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_0/width_65k/average_l0_72

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.0.hook_mlp_out

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

↵

-0.69

invokeLater

-0.62

SequentialGroup

-0.61

de

-0.59

De

-0.56

des

-0.55

 involves

-0.53

ц

-0.53

Des

-0.52

lara

-0.52

POSITIVE LOGITS

 sanitaires

0.93

horabuena

0.90

 avoient

0.85

 femininos

0.85

Autoritní

0.85

 auroit

0.83

 ویکی‌پدیا

0.83

 espirituales

0.82

adə

0.82

 pouvoit

0.81

Activations Density 0.060%

Japanese particles and conjunctions indicating grammatical relationships

No Comments

No Known Activations