INDEX

Explanations

references related to scientific markers and testing results

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

exclusive words

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

Hauptartikel

-0.72

 kasarigan

-0.60

ロウィン

-0.60

ftagPool

-0.60

 Administrativna

-0.59

хьтан

-0.57

IntoConstraints

-0.55

GEBURTSDATUM

-0.55

makeConstraints

-0.55

 ſeveral

-0.54

POSITIVE LOGITS

 naman

0.48

 then

0.36

Tē

0.35

 Casa

0.35

 Abre

0.33

 Hide

0.33

Casa

0.33

 Constant

0.32

 hide

0.32

Activations Density 0.295%

references related to scientific markers and testing results

exclusive words

No Comments

No Known Activations