INDEX

Explanations

mathematical operators and references to quantitative data

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

cut, reduction, greater, then

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

 Flach

-0.41

 Klassi

-0.40

тным

-0.40

 nuanced

-0.38

 Fried

-0.38

 pâtes

-0.38

 Spez

-0.37

 rarity

-0.37

 बह

-0.37

 Regel

-0.37

POSITIVE LOGITS

:✨

0.70

 يتيمه

0.69

 ffilmiau

0.61

 queſta

0.60

principalColumn

0.59

OGND

0.59

 незавершена

0.58

 Wikimedijinoj

0.57

 estekak

0.57

 BoxFit

0.56

Activations Density 0.004%

mathematical operators and references to quantitative data

cut, reduction, greater, then

No Comments

No Known Activations