INDEX

Explanations

references to proposed methods and their effectiveness in research

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

proposed method or approach

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

 nakalista

-0.65

hoeddwyd

-0.55

moveToFirst

-0.54

 Италијани

-0.52

Jeux

-0.52

хьтан

-0.52

 noDo

-0.50

省市镇

-0.50

󠁴

-0.50

orianCalendar

-0.48

POSITIVE LOGITS

 approach

0.39

 approaches

0.34

szön

0.32

 abordagem

0.32

 excepto

0.31

 surla

0.31

intz

0.31

 său

0.31

 method

0.31

 terbukti

0.31

Activations Density 0.081%

references to proposed methods and their effectiveness in research

proposed method or approach

No Comments

No Known Activations