finoptix14b / README.md
ccortezb's picture
fix: restore sdk_version 6.18.0 (gr.Server needs Gradio 6+)
28e7c6c verified
|
Raw
History Blame Contribute Delete
1.77 kB
---
title: FinOptix-14B
emoji: πŸ’°
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 6.18.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: Your FinOps architect in 14 billion parameters
---
# FinOptix-14B β€” FinOps Command Center
[![Demo Video](https://img.shields.io/badge/Demo-YouTube-red)](https://www.youtube.com/watch?v=9qMdzZVqS2Q)
[![Model](https://img.shields.io/badge/Model-HuggingFace-yellow)](https://huggingface.co/ccortezb/FinOptix-14B)
[![GGUF](https://img.shields.io/badge/GGUF-Q4__K__M-blue)](https://huggingface.co/ccortezb/FinOptix-14B-GGUF)
An autonomous FinOps specialist running on open weights (Qwen 2.5 14B + QLoRA). Trained for **$3.50**.
## 🎬 Demo Video
https://www.youtube.com/watch?v=9qMdzZVqS2Q
## Capabilities
- **Terraform HCL Auditor** β€” Rightsizing, tagging, governance findings + optimized code
- **Cost Parser** β€” AWS cost/anomaly JSON analysis with financial metrics
- **BYaML Governance** β€” Architecture validation against policies and catalog
- **Ask FinOptix** β€” Free-form FinOps Q&A with actionable recommendations
## Run Locally (GGUF)
```bash
ollama run hf.co/ccortezb/FinOptix-14B-GGUF
```
## Training Details
| Spec | Value |
|------|-------|
| Base | Qwen 2.5 14B Instruct |
| Method | QLoRA (r=16, Ξ±=32) |
| Dataset | 265 gold examples |
| Hardware | Modal A100-80GB |
| Time | 40 minutes |
| Cost | $3.50 |
## Badges Earned (6/6)
βœ… Well-Tuned Β· βœ… Off-Brand Β· βœ… Field Notes Β· βœ… Sharing is Caring Β· βœ… Llama Champion Β· βœ… Off the Grid
## Author
**Carlos Cortez** β€” AWS Community Hero, CTO @ BWIT
[LinkedIn](https://linkedin.com/in/carloscortezcloud) Β· [X](https://x.com/ccortezb) Β· [Blog](https://hf.co/blog/ccortezb/finoptix-14b-finops-slm)