INDEX

Explanations

elements related to humor and weekly activities

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

day of week posts

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

省市镇

-0.53

 INTERESAR

-0.51

Παραπομπές

-0.48

Manbalar

-0.47

 kaarangay

-0.45

angliski

-0.45

Autoritní

-0.44

RTLD

-0.43

 Normdatei

-0.41

ͧ

-0.41

POSITIVE LOGITS

 weekly

0.48

ValueStyle

0.47

 EconPapers

0.47

IVEREF

0.46

 podcast

0.43

DayOfWeek

0.41

 improvis

0.40

 enca

0.40

reihe

0.39

Efq

0.39

Activations Density 0.025%

elements related to humor and weekly activities

day of week posts

No Comments

No Known Activations