Upload README.md with huggingface_hub

README.md

@@ -1,5 +1,23 @@
+---
+license: other
+language:
+  - en
+pipeline_tag: text-generation
+library_name: mlx
+tags:
+  - mlx
+  - svg
+  - code-generation
+  - graphics
+  - vector-graphics
+  - apple-silicon
+  - llama
+---
+
 # SVG-Master
 
+**Subtitle**: Expert SVG generation from natural language.
+
 SVG-Master is a specialized fine-tuned LLM designed for high-quality **SVG code generation**. It takes natural language descriptions and outputs clean, valid, and optimized SVG code for icons, illustrations, and graphics.
 
 Built on top of **Llama 3.2 3B Instruct**, this model has been instruction-tuned specifically to understand geometry, color theory, and SVG syntax.
@@ -43,11 +61,8 @@ python -m mlx_lm.generate --model ./weights --adapter-path ./adapters --prompt "
 You can run the dequantized/fused version of the model directly in Ollama.
 
 ```bash
-# Create the model in Ollama
-ollama create svg-master -f Modelfile
-
-# Run the model
-ollama run svg-master "Generate a blue glowing circuit board icon"
+# Pull the model from Ollama Hub
+ollama run fahidnasir/svg-master "Generate a blue glowing circuit board icon"
 ```
 
 ### 3. Python API
@@ -64,17 +79,8 @@ print(response)
 
 - `weights/`: Base 4-bit model weights and configuration.
 - `adapters/`: LoRA adapter weights and training configuration.
-- `weight_fused/`: Fully fused, dequantized BF16 model (generated via Task 1).
+- `weight_fused/`: Fully fused, dequantized BF16 model.
 - `data/`: Prepared training/validation datasets.
-- `Modelfile`: Configuration for Ollama deployment.
-
-## 🔄 Dequantization & Fusion
-
-To recreate the original sized (FP16/BF16) model from the 4-bit weights and adapters:
-
-```bash
-uv run mlx_lm.fuse --model ./weights --adapter-path ./adapters --save-path ./weight_fused --dequantize
-```
 
 ## 📜 License
 
@@ -84,3 +90,4 @@ This model inherits the [Llama 3.2 Community License Agreement](https://www.llam
 
 - **HuggingFace**: [fahidnasir/SVG-Master](https://huggingface.co/fahidnasir/SVG-Master)
 - **Kaggle**: [nfahid/svg-master](https://www.kaggle.com/models/nfahid/svg-master)
+- **Ollama**: [fahidnasir/svg-master](https://ollama.com/fahidnasir/svg-master)