INDEX

Explanations

various formatting elements related to brackets and delimiters

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

multi-language words

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

 ویکی‌پدی

-0.75

 Chwiliwch

-0.74

 BoxFit

-0.67

دانشنامهٔ

-0.63

ArrowToggle

-0.61

تقاوى

-0.61

transQ

-0.60

withIdentifier

-0.60

XmlRootElement

-0.60

msgSender

-0.59

POSITIVE LOGITS

esistenza

0.40

esterno

0.39

 kebenaran

0.39

 wnios

0.38

appartamento

0.38

 concernés

0.38

 floresta

0.38

 fédé

0.37

 Genehmigung

0.36

 合集

0.35

Activations Density 0.001%

various formatting elements related to brackets and delimiters

multi-language words

No Comments

No Known Activations