Hugging Face's logo

EnderArcane-Studio
/
PCoreX-Turkish-Model

Safetensors
Turkish
llama
slm
turkish
language-model
wikipedia
smollm2
enderarcane-studio
Model card Files
xet
 Community

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this model content.

๐Ÿง  PCoreX V1 - Turkish SLM (Small Language Model) Demo

PCoreX V1 is a Small Language Model (SLM) pre-trained and fine-tuned on a massive Turkish general knowledge corpus. This Gradio space is an interactive demo developed by EnderArcane Studio to test the model's text completion capabilities and Turkish language structure.

โš ๏ธ Attention

  • This model was created by fine-tuning and making additions to SmolLM2-135M.
  • The training data includes the Turkish Alpaca dataset and exactly 20,893,891 lines of open-source Turkish Wikipedia data.

๐Ÿ“Œ Model Overview & Training Details

  • Model Version: PCoreX V1
  • Model Type: SLM (Small Language Model)
  • Base Architecture: SmolLM2-135M (Apache 2.0 Licensed)
  • Developer: EnderArcane Studio
  • Datasets Used:
    • Turkish Alpaca Dataset
    • Open-Source Turkish Wikipedia Dataset (~20,893,891 lines)
  • License: Apache License 2.0

โš ๏ธ Current Capabilities & Limitations

Note: PCoreX V1 is currently a raw pre-trained base model with initial fine-tuning.

  • It has not fully undergone conversational instruction tuning, which means it cannot hold a natural chat conversation or reliably follow complex user commands.
  • Its primary focus is text completion, understanding Turkish language patterns, and utilizing general knowledge acquired from over 20 million lines of Wikipedia and instruction data.

๐Ÿ› ๏ธ How It Works & How to Test

Since the model is primarily focused on text completion, it predicts and completes the continuation of a given Turkish prompt.

Example Usage (Text Completion):

  • Input (Prompt): Tรผrkiye'nin baลŸkenti Ankara,
  • Expected Output (Completion): ...iรง anadolu bรถlgesinde yer alan ve sanayi, eฤŸitim, kรผltรผr aรงฤฑsฤฑndan geliลŸmiลŸ bir ลŸehirdir.

๐Ÿš€ Roadmap

  • PCoreX V1.5 / V2: Advanced instruction tuning and full chat/dialogue capabilities.
  • PCoreX Music Edition: Integrate music generation, auto-loop creation, and audio processing modules.
  • Hugging Face Model Hub Release: Open-sourcing model weights for public access.

๐Ÿ“œ License & Attribution

This project is licensed under the Apache License 2.0.

  • Base Model: SmolLM2-135M (Licensed under Apache 2.0).
  • Modifications & Turkish Fine-tuning: Developed and maintained by EnderArcane Studio.

๐Ÿข Developer Info

Downloads last month
19
Safetensors
Model size
0.1B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Spaces using EnderArcane-Studio/PCoreX-Turkish-Model 2