INDEX

Explanations

specific brand names and high-performance components in various contexts

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

Mercury, Zeiss, Dana, brands

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

filepath

-0.35

<?

-0.35

XmlAccessorType

-0.35

 halben

-0.34

Jegyzetek

-0.32

endpush

-0.32

็ม

-0.31

 gustado

-0.31

 Scherer

-0.31

 cancha

-0.31

POSITIVE LOGITS

تقاوى

0.66

libft

0.62

RTLI

0.57

AnimationsModule

0.57

 LayoutInflater

0.56

 AspNetCore

0.55

branded

0.52

MethodManager

0.52

Personensuche

0.52

béco

0.50

Activations Density 0.022%

specific brand names and high-performance components in various contexts

Mercury, Zeiss, Dana, brands

No Comments

No Known Activations