INDEX

Explanations

references to specific names or titles

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

Tokens appearing inside parentheses or quotation marks

np_acts-logits-general · gemini-2.0-flash

nicknames or abbreviations in parentheses

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 Versuch

-0.44

-------------</

-0.42

melding

-0.40

 تضيفلها

-0.40

 voorzien

-0.37

rollo

-0.36

olem

-0.35

 Sonata

-0.35

istible

-0.35

 tijden

-0.34

POSITIVE LOGITS

 resourceCulture

0.69

 CURIAM

0.54

 nickname

0.54

 intptr

0.48

 affectionate

0.47

borderSide

0.47

kwds

0.46

affection

0.46

AccessorTable

0.46

綽

0.46

Activations Density 0.126%

references to specific names or titles

Tokens appearing inside parentheses or quotation marks

nicknames or abbreviations in parentheses

No Comments

No Known Activations