Upload model_readme.md with huggingface_hub

model_readme.md

@@ -0,0 +1,80 @@
+# DISARM Election Watch - Fine-tuned Llama-3.1 Model
+
+## Model Description
+
+This is a fine-tuned version of the Llama-3.1 model specifically optimized for DISARM Framework analysis of election-related content. The model has been trained on a comprehensive dataset of Nigerian election content from multiple platforms.
+
+## Model Details
+
+- **Base Model**: ArapCheruiyot/disarm_ew-llama3
+- **Fine-tuning Method**: LoRA (Low-Rank Adaptation)
+- **Optimization**: Apple Silicon (M1 Max) optimized
+- **Training Data**: 6,019 examples from multiple sources
+- **Task**: DISARM Framework classification and narrative analysis
+
+## Training Configuration
+
+- **LoRA Rank**: 16
+- **Batch Size**: 1
+- **Learning Rate**: 3e-4
+- **Sequence Length**: 2048
+- **Training Iterations**: 600
+- **Final Training Loss**: 1.064
+- **Final Validation Loss**: 1.354
+
+## Usage
+
+### With MLX-LM (Fused Model)
+```python
+from mlx_lm import load, generate
+
+model, tokenizer = load("ArapCheruiyot/disarm-ew-llama3-finetuned")
+# Use the model for inference
+```
+
+### With LoRA Adapters
+```python
+from mlx_lm import load, generate
+
+model, tokenizer = load("ArapCheruiyot/disarm_ew-llama3", 
+                        adapter_path="ArapCheruiyot/disarm-ew-llama3-finetuned")
+```
+
+### Example Usage
+```python
+prompt = """### Instruction:
+Classify the following content according to DISARM Framework techniques and meta-narratives:
+
+### Input:
+A viral WhatsApp broadcast claims that the BVAS machines have been pre-loaded with votes by INEC in favour of the incumbent party.
+
+### Response:"""
+
+response = generate(model, tokenizer, prompt, max_tokens=256, temp=0.1)
+print(response)
+```
+
+## Dataset
+
+The model was trained on the DISARM Election Watch dataset, which contains:
+- 6,019 examples from multiple platforms
+- DISARM Framework classifications
+- Meta-narrative analysis
+- Multi-language support
+
+## Performance
+
+Optimized for Apple Silicon with:
+- Metal GPU acceleration
+- 16GB wired memory optimization
+- Native MLX-LM framework
+
+## Model Files
+
+- **Fused Model**: Complete fine-tuned model (16GB)
+- **LoRA Adapters**: Lightweight adapters for efficient loading
+- **Checkpoints**: Training checkpoints every 100 iterations
+
+## License
+
+This model is part of the DISARM Election Watch project.