LocalAI-boost/localai-boost-starter

🚀 LocalAI Boost Pack — Free Starter Edition

Turn Generic Local Models into High-Performance Specialty AI Tools
Tailored for Ollama, LM Studio, and Open WebUI.

---

⚡ Why Local LLMs Underperform Out-of-the-Box

When you download a raw 7B or 8B model (like Llama 3.1 8B or Mistral 7B), you often encounter frustrating bottlenecks:

1. Endless "Yap" & Disclaimers: "Certainly! Here is an explanation of what I am about to do..." wasting precious generation tokens and time.
2. Uncalibrated Temperatures: High temperatures in coding prompts cause subtle syntax bugs; low temperatures in creative prompts make characters robotic and dull.
3. Truncated Contexts: Default 2048 context lengths crash multi-turn conversations and script refactoring.

LocalAI Boost Pack fixes this immediately with battle-tested system prompts featuring strict Negative Constraints, custom inference parameters (temperature, top_p, repeat_penalty), and an expanded 8,192 context window.

---

📦 What's Inside the Free Starter Pack

This free starter edition includes 4 essential tuned engines:

Preset Name	Focus Area	Tuned Behavior & Output Format
★ no-yap-coder	Dev / Coding	Zero polite talk, zero preamble. Immediately outputs executable, production-grade code blocks with correct syntax tags.
★ pc-auto-scripter	Automation	Specializes in Windows Batch (.bat), PowerShell (.ps1), and Bash. Includes safety checks and milestone echoes.
★ raw-brainstormer	Creative / Strategy	Cuts through generic corporate platitudes. Delivers 10 punchy, high-impact, provocative concepts in a numbered list.
★ character-soul-builder	Creative / Fiction	Builds multidimensional characters with flaws, contradictions, sensory quirks, and high-stakes dialogue samples.

---

🛠️ Quick Installation Guide

Option 1: Ollama (1-Click Automated Setup)

Windows

1. Make sure Ollama is installed and running.
2. Double-click install_windows.bat.
3. The script will automatically pull llama3.1:8b (if needed) and register all 4 custom models.
4. Open your command prompt / terminal and run:

   ollama run no-yap-coder
   

Mac / Linux

1. Open terminal and navigate to this folder.
2. Make the script executable and run:

   chmod +x install_mac_linux.sh
   ./install_mac_linux.sh
   

3. Run any model:

   ollama run raw-brainstormer
   

---

Option 2: LM Studio Setup

1. Launch LM Studio.
2. Load your desired model (e.g., Meta-Llama-3.1-8B-Instruct or Qwen2.5-7B-Instruct).
3. Open the JSON_Presets folder in this pack.
4. Copy the system_prompt from the corresponding .json file into LM Studio's System Prompt field.
5. In LM Studio's right-hand sidebar under Inference Parameters, apply the values from the JSON:
   • Temperature: 0.2 (for dev) or 0.8–0.85 (for creative)
   • Top P: 0.85 or 0.95
   • Repeat Penalty: 1.1 or 1.2
   • Context Length: 8192

---

Option 3: Open WebUI Setup

1. Go to Workspace > Models in Open WebUI.
2. Click Add Model (+).
3. Paste the system_prompt and configure parameters from JSON_Presets/<name>.json.
4. Click Save & Create.

---

🔄 How to Change the Base Model

By default, all Modelfiles reference FROM llama3.1:8b. If you prefer using other weights:

1. Open any .Modelfile in a text editor (Notepad, VS Code).
2. Edit the first line to your preferred local model tag:

   # Example: Switch to Qwen 2.5 Coder
   FROM qwen2.5-coder:7b
   
   # Example: Switch to Mistral
   FROM mistral:7b
   

3. Re-run ollama create <name> -f <path> to update.

---

💎 Unlock the Full Arsenal: LocalAI Boost Pack PRO

Love the starter pack? Elevate your entire local AI workflow with the PRO Complete Edition:

🔥 What You Get in PRO:

• All 20 Production-Calibrated Engines across 2 comprehensive categories:
  • 10 Dev & Automation Engines: no-yap-coder, pc-auto-scripter, bug-sniper, cli-command-gen, code-refactor-pro, json-data-converter, api-tool-architect, sql-query-builder, doc-readme-writer, tech-eli5-tutor.
  • 10 Creative & Ideation Engines: raw-brainstormer, character-soul-builder, world-lore-architect, plot-twist-engine, dialogue-naturalizer, dark-gritty-writer, devil-advocate-critic, visual-prompt-director, viral-hook-creator, infinite-rp-master.
• In-Depth Parameter Tuning Cheatsheet (PARAMETER_CHEATSHEET.md): Learn the exact mathematical sweet spots for temperature, repeat penalties, and VRAM context sizing.
• 1-Click Master Installer for both Windows and Mac/Linux.
• Lifetime Updates & Future Preset Additions.

👉 Get the Pro Complete Pack on Gumroad
(Use code BOOST20 at checkout for an instant 20% launch discount!)

---

📄 License & Community

• Free for personal and commercial use on your own local machines.
• Created for the local AI and open-weights developer community.