README.md

---
title: Coding Agent
emoji: 🏃
colorFrom: yellow
colorTo: green
sdk: docker
pinned: false
license: apache-2.0
---


---
title: Coding Agent Qwen-SFT v3
emoji: 🚀
colorFrom: blue
colorTo: gray
sdk: docker
pinned: false
license: apache-2.0
---

# 🚀 Coding Agent Qwen-SFT v2 (GGUF)

This is a fine-tuned **Qwen2.5-Coder-7B** model, optimized for production-grade Python, C++, and Rust development. 

### 🌟 Key Features
- **Logic-First:** Trained to avoid "Textbook" descriptions and provide direct code solutions.
- **Efficient:** Quantized to 4-bit (GGUF) to run smoothly on standard CPUs.
- **Context-Aware:** Handles complex concurrency, memory safety, and architectural prompts.

### 💻 Use it Locally (Ollama)
You can run this exact model on your own machine:
1. **Install [Ollama](https://ollama.com)**
2. **Download the GGUF file** from this repository.
3. **Create a `Modelfile`**:
   ```dockerfile
   FROM ./coding-agent-qwen-sft-v2-GGUF.q4_k_m.gguf
   PARAMETER temperature 0.4
   PARAMETER repeat_penalty 1.2
   ```
4. **Run:** `ollama create coding-agent -f Modelfile` then `ollama run coding-agent`

### 🛠️ Inference Specs
- **Repetition Penalty:** 1.2
- **Temperature:** 0.4
- **Stop Token:** `<|im_end|>`


Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference