EnderArcane

community

Activity Feed

AI & ML interests

None defined yet.

Organization Card

Community About org cards

⭐ EnderArcane Studio

AI • Software • Creative Technology • Open Source

EnderArcane Studio is an independent technology and development studio focused on building open-source software, artificial intelligence systems, datasets, developer tools, and creative technologies.

Our projects range from small language models and AI datasets to experimental applications, automation systems, music technology, and developer infrastructure.

🚀 What We Build

EnderArcane Studio works across several areas:

🤖 Artificial Intelligence & Machine Learning
🧠 Small Language Models (SLMs)
📚 Open Datasets & Data Processing
💻 Developer Tools & Software
🎵 Music & Audio AI
⚙️ Automation & AI Agents
🌐 Web Applications & Infrastructure
🧪 Experimental & Research Projects

The goal is not simply to build applications, but to experiment with the technologies behind them and make useful parts of that work accessible to others.

🧠 PCoreX

One of the studio's main AI projects is PCoreX, a family of experimental small language models and AI systems.

PCoreX currently focuses heavily on Turkish language processing and efficient small-scale models.

PCoreX V1

PCoreX V1 is a Turkish SLM based on SmolLM2-135M.

Its training work includes Turkish instruction data and a large open-source Turkish Wikipedia corpus.

The current V1 primarily focuses on:

Turkish language structure
Text completion
General knowledge
Efficient small-model experimentation
Further model development

PCoreX V1 is considered an early-stage model rather than a fully instruction-tuned conversational assistant.

PCoreX Roadmap

Future PCoreX development includes:

Improved instruction tuning
Better conversational capabilities
PCoreX V2 and later versions
Specialized model variants
Music and audio-related AI systems
More Turkish training data
Improved training and evaluation pipelines

📚 Datasets

EnderArcane Studio also develops and maintains datasets used for AI research and model training.

PCoreX-TR-Alpaca

A Turkish instruction dataset prepared for PCoreX development.

PCoreX-TR-Alpaca-V2

A larger processed dataset containing 623,978 records combining multiple Turkish-language sources.

The dataset is structured around fields such as:

text
source
type

It is intended primarily for Turkish language-model and instruction-tuning research.

🎵 UserLooperGen & PCoreX Music Edition

EnderArcane Studio also works on the intersection of music creation and artificial intelligence.

UserLooperGen is a music-loop generation and processing project.

Its planned AI ecosystem includes PCoreX Music Edition, with research into:

Automatic loop generation
Audio analysis
Music-loop processing
Audio feature extraction
AI-assisted music workflows

🧪 Experimental AI & Automation

Not every project at EnderArcane Studio is a finished product.

Some projects exist specifically to test:

AI agents
Automated development workflows
Dataset pipelines
Model training systems
AI-assisted programming
Cloud-based development
Hugging Face Spaces
API-based automation

These experiments often become the foundation for larger projects later.

🛠️ Technology

Our projects may involve technologies such as:

Python
JavaScript
HTML / CSS
PyTorch
Hugging Face Transformers
Hugging Face Datasets
Gradio
Cloud-based GPU environments
GitHub
Cloudflare
Google Apps Script

The exact technology stack varies between projects.

🌐 Hugging Face

Hugging Face is one of the main platforms used by EnderArcane Studio for publishing and experimenting with:

Models
Datasets
Spaces
Training experiments
AI demonstrations

🏗️ Project Organization

As the studio grew, some projects became large enough to deserve their own dedicated development organizations.

One example is HtStudio, which has its own organization because its growing collection of tools and projects would otherwise make the main EnderArcane Studio organization unnecessarily crowded.

This allows EnderArcane Studio to remain focused on its core AI and experimental projects while specialized ecosystems can evolve independently.