novaforgeai/ starcoder2:3b-optimized

153 1 month ago

NovaForge AI – StarCoder2-3B Optimized is a fast, CPU-only coding assistant built for accurate code generation.Optimized for low-RAM systems, it generates clean, efficient, and best-practice code across multiple programming languages — fully offline. 

ollama run novaforgeai/starcoder2:3b-optimized

Details

1 month ago

1d57176fabde · 1.7GB ·

model
starcoder2
·
3.03B
·
Q4_0
license
BigCode Open RAIL-M v1 License Agreement Section I: Preamble This OpenRAIL-M License Agreement was c
template
<file_sep> {{- if .Suffix }}<fim_prefix> {{ .Prompt }}<fim_suffix>{{ .Suffix }}<fim_middle> {{- else
system
You are NovaForge AI Code Assistant - specialized in writing, explaining, and debugging code. Focus
params
{ "num_batch": 128, "num_ctx": 2048, "num_gpu": 0, "num_predict": 1024, "num_thr

Readme

NovaForge AI – StarCoder2-3B Optimized

A lightweight, CPU-optimized code generation model based on StarCoder2-3B, tuned specifically for writing, debugging, and refactoring high-quality code on local machines.

🚀 Key Features

💻 Specialized for code generation

⚡ Very fast CPU inference

🧠 Accurate algorithms & logic

📄 Clean, readable, best-practice code

🔒 Runs fully offline for privacy

📦 Model Details

Model Name: novaforgeai/starcoder2:3b-optimized

Base Model: StarCoder2-3B

Model Size: ~1.7 GB

RAM Usage: ~2.2 GB

Context Length: 2048 tokens

Device: CPU-only (No GPU required)

🎯 Best Use Cases

Code generation (Python, JS, C++, Java, etc.)

Algorithms & data structures

Bug fixing & debugging

Code refactoring

Technical documentation

Best-practice examples

⚙️ Optimization Highlights

Low temperature for precise code

Reduced repetition

Longer output for full implementations

Tuned for accuracy over creativity

📥 Installation ollama pull novaforgeai/starcoder2:3b-optimized

▶️ Usage Interactive Coding ollama run novaforgeai/starcoder2:3b-optimized

Generate Code ollama run novaforgeai/starcoder2:3b-optimized “Write a Python function for binary search”

💻 System Requirements Minimum

CPU: 4 cores

RAM: 8 GB

Storage: 2 GB

OS: Windows / Linux / macOS

Recommended

CPU: 6+ cores

RAM: 16 GB

SSD storage

📊 Performance Summary

Speed: Very fast (3–8s typical)

Code Quality: Clean & structured

Best For: Full code generation

Not Ideal For: Casual chat or theory discussion

👉 Tip: Use Qwen 2.5-3B Optimized for code explanations or discussions.

🔐 Privacy

Runs entirely on your local machine. No cloud calls, no tracking, no data sharing.

📄 License

Based on StarCoder2 Original license applies — review before commercial use.

🤝 Credits

Base Model: StarCoder2 Team

Optimization & Packaging: NovaForge AI Team

Tools: llama.cpp, Ollama

🔗 Related Models

General Purpose: novaforgeai/qwen2.5:3b-optimized

Ultra-Fast: novaforgeai/gemma2:2b-optimized

Advanced Coding: novaforgeai/novaforge-mistral:7b-q4km

Maintained by: NovaForge AI Team Status: ✅ Production Ready Specialization: 💻 Code Generation