INDEX

Explanations

characteristic elements or symbols related to structured data or coding

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

 prohibido

-0.51

 béisbol

-0.49

 näky

-0.48

 immédi

-0.47

 tämän

-0.47

 aikaa

-0.46

 expuesto

-0.46

 kautta

-0.46

 picioare

-0.44

 myö

-0.44

POSITIVE LOGITS

SwitchCompat

0.57

 nonUne

0.55

EndInit

0.55

 ویکی‌پدی

0.53

 TextAppearance

0.52

 useContext

0.52

InitStruct

0.51

PhysRevD

0.50

PageModule

0.50

ierr

0.50

Activations Density 0.335%

characteristic elements or symbols related to structured data or coding

No Comments

No Known Activations