INDEX

Explanations

references to academic citations and publications

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

academic journals and citations

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Smiling

-0.35

smile

-0.34

RRect

-0.33

ady

-0.32

 smiles

-0.30

 flotte

-0.30

metti

-0.30

 dead

-0.30

 child

-0.30

foot

-0.29

POSITIVE LOGITS

 ویکی‌پدی

0.84

GEBURTSDATUM

0.65

 Biôgrafia

0.61

 Савезне

0.60

دانشنامهٔ

0.53

 Italijanski

0.52

 informée

0.51

 defaultstate

0.51

">//

0.50

 Италијани

0.49

Activations Density 0.234%

references to academic citations and publications

academic journals and citations

No Comments

No Known Activations