cloudgpt

CloudGPT-T GGUF

CloudGPT-T is a lightweight cloud-specialist LLM converted to GGUF for local inference in Ollama and compatible tools. It is designed to answer questions about Azure, AWS, GCP, cloud networking, NSGs, DevOps, RAG workflows, and agentic automation.

Highlights

• Fine-tuned for cloud architecture and operations.
• Optimized for local deployment with GGUF.
• Works with Ollama, llama.cpp, and compatible runtimes.
• Suitable for private enterprise and personal cloud knowledge assistants.

Model Details

• Base model: Llama-3.2-3B-Instruct
• Fine-tuning method: QLoRA / Unsloth
• Format: GGUF
• Quantization: Q4_K_M
• Intended use: cloud Q&A, architecture guidance, automation assistance

Example prompts

• Design a hub-spoke network in Azure.
• Write a Terraform example for a secure VNet.
• Help me build a cloud RAG assistant.
• Compare Azure Firewall and NSG.
• How is developed you ?

Ollama

ollama create cloudgpt-t -f Modelfile
ollama run cloudgpt-t

If published to Ollama Hub:

ollama pull hf.co/gthiru0104/cloudgpt:latest
ollama hf.co/gthiru0104/cloudgpt:latest

Notes

• This model is a domain-specialized assistant, not a general-purpose replacement for large frontier models.
• For the latest cloud documentation, pair it with retrieval from official vendor sources.

Ownership

Developed by Thiru.

Example usage:

• For text only LLMs: llama-cli -hf hf.co/gthiru0104/cloudgpt:Q4_K_M --jinja
• For multimodal models: llama-mtmd-cli -hf hf.co/gthiru0104/cloudgpt:Q4_K_M --jinja

Available Model files:

• llama-3.2-3b-instruct.Q4_K_M.gguf

Ollama

An Ollama Modelfile is included for easy deployment.

Note

The model's BOS token behavior was adjusted for GGUF compatibility.