INDEX

Explanations

user interaction and user-driven actions related to software or applications

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

Actions like reading, viewing, listening, or accessing

np_acts-logits-general · gemini-2.0-flash

browse, read, find, listen

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



-0.68

 propOrder

-0.54

ftagPool

-0.51

 mxArray

-0.47

Tembelea

-0.45

medriver

-0.44

SerializeField

-0.44

 GenerationType

-0.43

GEBURTSDATUM

-0.42

тистика

-0.42

POSITIVE LOGITS

 الاطلاع

0.53

 اطلع

0.41

 Eindruck

0.40

Accessors

0.39

комиться

0.36

知る

0.35

 browsing

0.35

ôtel

0.35

 accessing

0.34

ContentAsync

0.34

Activations Density 0.412%

user interaction and user-driven actions related to software or applications

Actions like reading, viewing, listening, or accessing

browse, read, find, listen

No Comments

No Known Activations