Improve model card: link paper and code, update license, and add tags

README.md

@@ -1,112 +1,72 @@
----
-library_name: peft
-license: other
-base_model: Qwen/Qwen3-8B
-tags:
-- base_model:Qwen/Qwen3-8B
-- llama-factory
-- lora
-- transformers
-pipeline_tag: text-generation
-model-index:
-- name: sft100
-  results: []
----
-
-
-# GlobalHealthAtlas Public Model
-
-This repository contains the code and resources for the GlobalHealthAtlas Public Model project. The project hosts a fine-tuned question-answering model designed to provide helpful answers to public health–related queries.
-
-## Overview
-
-The GlobalHealthAtlas Public Model is designed to provide informative, context-aware answers to questions about public health documents and topics. It includes the fine-tuned adapter model, tokenizer files, example prompts, and scripts for running inference in a Q&A or conversational setting.
-
-## Files and Directories
-
-- **adapter_config.json**: Configuration for the adapter model.
-- **adapter_model.safetensors**: Pre-trained adapter model.
-- **added_tokens.json**: Tokens added to the tokenizer.
-- **all_results.json**: Example outputs and sample Q&A responses from inference and evaluation.
-- **merges.txt**: Merges for the tokenizer.
-- **README.md**: This file.
-- **special_tokens_map.json**: Mapping of special tokens.
-- **tokenizer_config.json**: Configuration for the tokenizer.
-- **tokenizer.json**: Tokenizer file.
-- **train_results.json**: Results from training.
-- **trainer_log.jsonl**: Log of the training process.
-- **trainer_state.json**: State of the trainer.
-- **checkpoint-990/**: Directory containing checkpoint files for the model.
-
-## Usage
-
-To use the GlobalHealthAtlas Public Model, clone the repository and follow the instructions in the README files. Ensure that all dependencies are installed and configured correctly. Load `adapter_model.safetensors` and the tokenizer, then run the provided inference scripts or your own prompt loop to perform question-answering. This model is intended for research and information purposes and should not be used for clinical diagnosis or other high‑stakes decisions without human review.
-
-## License
-
-This project is licensed under the MIT License. See the LICENSE file for details.
-
-## Acknowledgments
-
-We would like to thank the contributors and the community for their support in developing this project.
-
-## References
-
-For more information, please refer to the related publications and documentation in the Public Health directory.
-
-## Model description
-
-This model is a fine-tuned version of [/root/Qwen3-8B](https://huggingface.co//root/Qwen3-8B) on the GlobalHealthAtlas dataset.
-
-
-## Intended uses & limitations
-
-This model is intended to be used as a question-answering and information retrieval component for the Global Health Atlas project and related public health queries. It is designed to provide helpful, evidence-informed answers but is NOT intended for clinical diagnosis, medical advice, or other high‑stakes decision-making. Users should validate performance on their own data prior to deployment, monitor for potential biases introduced by the fine-tuning data, and review outputs with domain experts when used for operational tasks.
-
-## Training and evaluation data
-
-The model was fine-tuned on the `GlobalHealthAtlas` dataset to adapt the Qwen/Qwen3-8B base model for question-answering and response generation. Evaluation used held-out splits and standard validation procedures. Detailed evaluation metrics, per-epoch results, and sample outputs are available in `train_results.json` and `all_results.json` in this repository. To reproduce results, use the same data splits and training hyperparameters listed below. For additional context about dataset preparation and task setup, consult the `Public_health/` directory and the Global Health Atlas repository: https://github.com/Jan8217/GlobalHealthAtlas.
-
-
-This model is a fine-tuned version of [/root/Qwen3-8B](https://huggingface.co//root/Qwen3-8B) on the GlobalHealthAtlas dataset
-
-## Model description
-
-More information needed
-
-## Intended uses & limitations
-
-More information needed
-
-## Training and evaluation data
-
-More information needed
-
-## Training procedure
-
-### Training hyperparameters
-
-The following hyperparameters were used during training:
-- learning_rate: 5e-05
-- train_batch_size: 1
-- eval_batch_size: 8
-- seed: 42
-- gradient_accumulation_steps: 8
-- total_train_batch_size: 8
-- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
-- lr_scheduler_type: cosine
-- lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 2.0
-
-### Training results
-
-Training completed for 2 epochs using the hyperparameters listed above. Final evaluation metrics (validation loss and task-specific scores) and per-epoch logs are stored in `train_results.json` and the training log is available in `trainer_log.jsonl`. Please refer to those files for exact numeric results and per-example breakdowns.
-
-
-### Framework versions
-
-- PEFT 0.15.1
-- Transformers 4.51.3
-- Pytorch 2.3.0+cu121
-- Datasets 3.2.0
-- Tokenizers 0.21.0
+---
+base_model: Qwen/Qwen3-8B
+library_name: peft
+license: mit
+pipeline_tag: text-generation
+tags:
+- base_model:Qwen/Qwen3-8B
+- llama-factory
+- lora
+- transformers
+- medical
+- public-health
+model-index:
+- name: sft100
+  results: []
+---
+
+# GlobalHealthAtlas Public Model
+
+This repository contains the fine-tuned LoRA adapter for the **GlobalHealthAtlas Public Model**, introduced in the paper [From Knowledge to Inference: Formalizing Specialized Public Health Reasoning on GlobalHealthAtlas](https://huggingface.co/papers/2602.00491).
+
+The model is designed to provide informative, context-aware answers to public health–related queries across 15 domains and 17 languages. It was fine-tuned from [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B) on the GlobalHealthAtlas dataset.
+
+- **Paper:** [From Knowledge to Inference: Formalizing Specialized Public Health Reasoning on GlobalHealthAtlas](https://huggingface.co/papers/2602.00491)
+- **Repository:** [Jan8217/GlobalHealthAtlas](https://github.com/Jan8217/GlobalHealthAtlas)
+
+## Overview
+
+Public health reasoning requires population-level inference grounded in scientific evidence, expert consensus, and safety constraints. This model addresses the scarcity of supervised signals in this domain. It was trained on the GlobalHealthAtlas dataset, a large-scale multilingual corpus of 280,210 instances spanning 15 domains including infectious disease prevention, health policy, and vaccination.
+
+## Intended Uses & Limitations
+
+This model is intended to be used as a question-answering and information retrieval component for research and public health queries. 
+
+**Important Disclaimer:** This model is **NOT** intended for clinical diagnosis, medical advice, or other high‑stakes decision-making without human review by domain experts. 
+
+## Training Procedure
+
+The model was fine-tuned using LoRA (Low-Rank Adaptation) on the Qwen3-8B base model via the [LLaMA Factory](https://github.com/hiyouga/LLaMA-Factory) framework.
+
+### Training Hyperparameters
+
+The following hyperparameters were used during training:
+- **Learning rate:** 5e-05
+- **Train batch size:** 1
+- **Gradient accumulation steps:** 8
+- **Total train batch size:** 8
+- **Optimizer:** AdamW
+- **LR scheduler type:** cosine
+- **LR scheduler warmup ratio:** 0.1
+- **Num epochs:** 2.0
+- **Seed:** 42
+
+### Framework Versions
+
+- PEFT 0.15.1
+- Transformers 4.51.3
+- Pytorch 2.3.0+cu121
+- Datasets 3.2.0
+- Tokenizers 0.21.0
+
+## Citation
+
+If you use this model or the GlobalHealthAtlas dataset, please cite:
+
+```bibtex
+@article{globalhealthatlas2026,
+  title={From Knowledge to Inference: Formalizing Specialized Public Health Reasoning on GlobalHealthAtlas},
+  author={GlobalHealthAtlas Team},
+  year={2026}
+}
+```