metadata
language:
- ta
- en
license: apache-2.0
library_name: llama.cpp
tags:
- tamil
- bilingual
- nlp
- classification
- crossberryweb
- gguf
extra_gated_heading: Tamila Model Access
extra_gated_button_content: Acknowledge
🚀 Tamila Master v0.3
Created by crossberryweb
Tamila is a high-performance bilingual model (Tamil/English) trained on a massive global corpus of over 2.2 million segments.
🔗 Project Links
- Live Demo (HF Space): https://huggingface.co/spaces/Crossberry/tamila-test-app
- Web Deployment: crossberry.vercel.app
- Dataset Repository: Hugging Face Tamila
📊 Model Benchmarks
| Task | Dataset | Accuracy | Loss |
|---|---|---|---|
| Global Corpus Tuning | 2.2M Segments | 1.0000 | 6.64e-10 |
| Literature (Thirukkural) | Kaggle NLP | 0.9868 | 0.0612 |
| Technical (Kimi K2) | PDF Extract | 1.0000 | 1.17e-06 |
🛠 Future Roadmap
- Integration with advanced Transformer architectures.
- Expanded support for regional Tamil dialects.
- Real-time API integration for mobile applications.
📖 More Info
This model utilizes a custom MLP architecture optimized for GGUF deployment. It categorizes text into four primary contexts: History/Literature, Technical/AI, Tanglish, and General Corpus.
Developed for the open-source community by Crossberryweb.