You can deploy this model and start earning money today!

HoangPhu/
ticket-classifier-qwen25-7b-gguf

Name: ticket-classifier-qwen25-7b-gguf
Availability: InStock

gguf

endpoints_compatible

conversational

Views

No views yet

IT Ticket Classifier - Qwen2.5 7B GGUF

Ready for Ollama! Direct conversion from original model.

Model Details

Base Model: unsloth/Qwen2.5-7B-Instruct
Fine-tuning: LoRA adapter merged
Format: GGUF (llama.cpp compatible)
Quantization: F16
Date Converted: 2026-06-08 03:53:34
Accuracy: 78.72%
Quality: 100% (original precision)

Quick Start

1. Clone Repository

bash

git clone https://huggingface.co/HoangPhu/ticket-classifier-qwen25-7b-gguf
cd ticket-classifier-qwen25-7b-gguf

2. Create Ollama Model

ollama create ticket-classifier -f Modelfile

3. Use Immediately

ollama run ticket-classifier "My VDI is not connecting"

File Sizes & Formats

Original (BF16): ~15 GB
GGUF F16: ~8 GB
Compression: 47%

Usage Examples

CLI

bash

ollama run ticket-classifier "Reset my password"
# Output: PASSWORD_RESET

Python API

python

import requests

url = "http://localhost:11434/api/generate"
payload = {
    "model": "ticket-classifier",
    "prompt": "Create new VDI",
    "stream": False,
    "temperature": 0.1,
}

response = requests.post(url, json=payload)
print(response.json()['response'])

Performance

Accuracy: 78.72%
Best Categories (F1 > 90%):
- Reset MFA O365: 97.2%
- Remove VDI: 97.1%
- Register Application Entra ID: 94.1%
- Enable VDI: 91.9%

System Requirements

RAM: 8+ GB (CPU), 12+ GB (GPU)
GPU: Optional but recommended (NVIDIA RTX 3060+)
Disk: 12 GB (model + cache)

Inference Speed

CPU: 5-10 sec/ticket
GPU (RTX 3080): 0.5-2 sec/ticket
GPU (RTX 4090): 0.2-0.5 sec/ticket

Format Information

GGUF Format:

✅ Ollama compatible
✅ No conversion needed
✅ Cross-platform (Windows/Mac/Linux)
✅ Optimized for inference
✅ Low memory footprint

F16 Quantization:

Quality retained: 100%
Size: ~8 GB
Speed: Excellent
Memory: Efficient

License

Apache 2.0 (same as Qwen2.5-7B-Instruct)