INDEX

Explanations

structural or formatting elements in text, especially related to classification or categorization

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

0h, 0l, half

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

 endIndex

-0.42

EndProject

-0.41

__(/*!

-0.40

Gears

-0.40

 ostavi

-0.39

 Hege

-0.39

 ApiException

-0.39

 мәкал

-0.38

skjaer

-0.38

󠁣

-0.38

POSITIVE LOGITS

ChildScrollView

0.64

+#+#

0.56

 metade

0.44

setContentView

0.43

 encarga

0.42

 referenties

0.41

Parcelize

0.41

Rüyada

0.40

gsSP

0.40

astéroïdes

0.40

Activations Density 0.059%

structural or formatting elements in text, especially related to classification or categorization

0h, 0l, half

No Comments

No Known Activations