INDEX

Explanations

numerical representations or patterns in text

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

3 followed by end of turn or $

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

 سكانية

-0.46

 تضيفلها

-0.44

ModelSerializer

-0.44

GEBURTS

-0.43

canestro

-0.39

醐

-0.36

 charme

-0.36

 Standardi

-0.36

 standardi

-0.36

//</

-0.35

POSITIVE LOGITS

 Administrativna

0.50

astéroïdes

0.49

GOTREF

0.48

complexContent

0.43

ثيق

0.43

 arşivlendi

0.42

RenderAtEndOf

0.42

 dépens

0.41

HasForeignKey

0.41

aarrggbb

0.40

Activations Density 0.195%

numerical representations or patterns in text

3 followed by end of turn or $

No Comments

No Known Activations