VoltageVagabond/spam-classifier-liquid-GGUF

Senior Project Notice

This repository was created for a senior project in ENGT 375 Applied Machine Learning at Old Dominion University. It is provided for educational and research demonstration purposes only. It is not intended for production use, security filtering, or making real-world spam/phishing decisions. Always use established security tools for operational email protection.

spam-classifier — GGUF

Educational Use Only Created as a senior capstone project for ENGT 375: Applied Machine Learning at Old Dominion University (Spring 2026). Not intended for production use.

A fully merged, standalone GGUF of LiquidAI/LFM2.5-1.2B-Instruct fine-tuned with LoRA to classify emails as SPAM, HAM, or PHISHING.

No separate adapter or base model needed — download and run directly.

---

Provided Files

Filename	Format	Size	Use case
spam-classifier-F16.gguf	GGUF F16	~2.2 GB	llama.cpp, LM Studio, Ollama

The system prompt is baked into the model's chat template — most clients (LM Studio, Ollama, llama.cpp via /v1/chat/completions) apply it automatically.

---

Model Details

Property	Value
Base model	LiquidAI/LFM2.5-1.2B-Instruct
Fine-tuning method	LoRA (rank 8, alpha 16)
Training task	3-class: SPAM / HAM / PHISHING
Training examples	~8,000 emails
Chat template	ChatML (<|im_start|> / <|im_end|>)
Recommended temperature	0.2
Max tokens	200

---

How to Download

huggingface-cli download VoltageVagabond/spam-classifier-liquid-GGUF \
    spam-classifier-F16.gguf \
    --local-dir ./spam-classifier-gguf

---

Usage

llama.cpp — Server

llama-server \
  -m spam-classifier-F16.gguf \
  --port 8081 \
  -ngl 99 \
  --temp 0.2 \
  --n-predict 200

Open http://127.0.0.1:8081 and send:

Classify this email as SPAM, HAM, or PHISHING. Give your classification on
the first line, then explain your reasoning in 2-3 sentences.

Email:
[paste email here]

llama.cpp — CLI

llama-cli \
  -m spam-classifier-F16.gguf \
  --temp 0.2 \
  -p "<|im_start|>system
You are an email spam classifier. Analyze the email and classify it as SPAM, HAM, or PHISHING. Explain your reasoning.<|im_end|>
<|im_start|>user
Classify this email as SPAM, HAM, or PHISHING. Give your classification on the first line, then explain your reasoning in 2-3 sentences.

Email:
Congratulations! You have won $1,000,000. Click here to claim your prize.<|im_end|>
<|im_start|>assistant
"

LM Studio

1. Download spam-classifier-F16.gguf
2. Open LM Studio → My Models → Import → select the file
3. The system prompt is pre-loaded from the model's chat template
4. Set Temperature to 0.2

Ollama

# Download the repo (includes Modelfile)
huggingface-cli download VoltageVagabond/spam-classifier-liquid-GGUF \
    --local-dir ./spam-classifier-gguf

# Create and run
ollama create spam-classifier -f ./spam-classifier-gguf/Modelfile
ollama run spam-classifier

---

Example Output

Input:

Classify this email as SPAM, HAM, or PHISHING.

Email:
Congratulations! You have won $1,000,000. Click here to claim your prize.

Output:

SPAM

This email exhibits classic spam indicators: an unrealistic prize claim,
urgency to "click here," and no sender context. The promise of $1,000,000
is a common social engineering tactic used to lure clicks.

---

Source

• Training code & adapter: VoltageVagabond/spam-classifier-liquid
• Training dataset: VoltageVagabond/spam-email-dataset
• Course: ENGT 375 — Applied Machine Learning, Old Dominion University, Spring 2026