INDEX

Explanations

elements related to structured data and programming

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

inbox filter, linear, consists of

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

 HasFactory

-0.50

 positively

-0.42

 barefoot

-0.40

шні

-0.39

pol

-0.39

 wear

-0.39

 ErrIntOverflow

-0.39

 aDecoder

-0.39

 Shetterly

-0.38

 fairly

-0.37

POSITIVE LOGITS

exitRule

0.55

'{@

0.53

 resourceCulture

0.51

 utafitiHapana

0.49

 שוליים

0.48

ValueStyle

0.48

calaure

0.47

adaptiveStyles

0.45

Autoritní

0.45

 Roskov

0.43

Activations Density 0.087%

elements related to structured data and programming

inbox filter, linear, consists of

No Comments

No Known Activations