INDEX

Explanations

conditional phrases related to recommendations and regulations

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

use, prescribing, or approving

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

Eti

-0.41

 dAtA

-0.41

NameInMap

-0.40

𝓛

-0.38

 Typhoon

-0.37

 nature

-0.37

Amo

-0.37

HtmlAttribute

-0.35

NUM

-0.35

Num

-0.34

POSITIVE LOGITS

 Keuangan

0.60

 mitään

0.56

 osobow

0.56

 traseiro

0.55

 isolado

0.54

 zijne

0.53

 financeiro

0.53

 esportivo

0.52

 eletrônico

0.52

 mijne

0.52

Activations Density 0.078%

conditional phrases related to recommendations and regulations

use, prescribing, or approving

No Comments

No Known Activations