📌 Model Summary Chakster/BreastCancerQA-TinyLlama is a fine-tuned version of the TinyLlama-1.1B-Chat model designed to provide accurate, concise, and context-aware answers to frequently asked questions related to breast cancer. It is specifically tailored for French-language Q&A use cases in medical applications such as health assistants and chatbots.
🩺 Intended Use This model is intended for:
Chatbots that provide breast cancer information to French-speaking users.
Educational tools for patient support and awareness.
Streamlit or web apps for multilingual health question-answering.
📂 Dataset Source: Custom dataset of 300+ French Q&A pairs on breast cancer.
Type: JSON format, questions and answers derived from verified medical resources like WHO, Cancer.net, Breastcancer.org, etc.
Topics Covered:
Symptoms
Diagnosis
Treatment (radiotherapy, surgery, chemotherapy, hormone therapy)
Prevention and screening
Risk factors
🧪 Training Details Framework: Hugging Face Transformers
Hardware: Google Colab with T4 GPU
Epochs: 3
Optimizer: AdamW
Evaluation Metrics:
ROUGE-L: ~0.85
BLEU-4: ~0.72
Perplexity: ~4.3
📊 Metrics Report (You can upload a .txt file if available)
Metric Value ROUGE-L 0.85 BLEU-4 0.72 Perplexity 4.3
⚠️ Limitations Not a replacement for professional medical advice.
May produce outdated or hallucinated answers if used outside the Q&A domain it was trained on.
Primarily effective for FAQ-style questions, not long or complex dialogues.
license: apache-2.0 datasets: - Chakster/breastcancer_QA_french.json language: - fr - en base_model: - TinyLlama/TinyLlama-1.1B-Chat-v1.0 pipeline_tag: text-generation
- Downloads last month
- -