INDEX

Explanations

patterns or identifiers in structured data or formatting

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

numbers and technical codes

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

 Turquía

-0.43

щадь

-0.38

 שוליים

-0.38

äsident

-0.38

 Moscú

-0.38

 Egipto

-0.38

 Irán

-0.38

 juge

-0.38

rhosis

-0.38

存于互联网档案馆

-0.37

POSITIVE LOGITS

󠁢

0.62

OGND

0.57

 própri

0.55

Diweddarwch

0.55

 îna

0.54

 North

0.52

Tämä

0.52

########.

0.51

 South

0.50

 European

0.49

Activations Density 0.001%

patterns or identifiers in structured data or formatting

numbers and technical codes

No Comments

No Known Activations