INDEX

Explanations

programming-related keywords and phrases, particularly those related to configurations and default settings

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

current state or category

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

 autorytatywna

-0.78

 مرئيه

-0.71

évaluateur

-0.71

niſſe

-0.68

IUrlHelper

-0.68

ⓧ

-0.67

KommentareTeilen

-0.66

 ujednoznacz

-0.66

<unused8>

-0.65

<unused14>

-0.65

POSITIVE LOGITS

 Jüng

0.34

 Steine

0.30

 Grä

0.29

 roja

0.29

 Utama

0.29

 ahora

0.28

 mucha

0.28

 amarilla

0.27

basicConfig

0.27

 benar

0.27

Activations Density 0.178%

programming-related keywords and phrases, particularly those related to configurations and default settings

current state or category

No Comments

No Known Activations