---
license: mit
datasets:
- cngchis/Support-Ticket-Router-12K-Cleaned
language:
- en
metrics:
- f1
- confusion_matrix
- precision
- recall
- accuracy
base_model:
- unsloth/Phi-4-mini-instruct
new_version: cngchis/phi4-mini-intent
pipeline_tag: text-classification
library_name: transformers
tags:
- phi
- nlp
- intent
- classification
- math
- code
- finance
---

## About

Static GGUF quantization for an **Intent Classification model**.

This model is converted from a Hugging Face checkpoint and optimized for local inference using `llama.cpp` compatible runtimes.

The model is designed for predicting intent labels from user input text.

---

## Usage

If you are unsure how to use GGUF files, refer to:
https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF

Basic usage with `llama.cpp`:

```bash
./main -m model.Q4_K_M.gguf -p "Your input text here"
```

For classification tasks, ensure your prompt format matches the training setup (e.g., instruction or label format).

---

## Provided Quant

|Link	|Type	    |Size/GB|Notes
|-------|-----------|-------|---------------------------------------|
|GGUF	|Q4_K_M	~X.X|2.5	|recommended balance of speed and quality

---

## Notes
- This model is fine-tuned for intent classification tasks only
- Best performance when input follows the same format as training data
- Q4_K_M provides a good trade-off between accuracy and inference speed

---

## FAQ / Requests
Thanks

Thanks to the open-source GGUF ecosystem (llama.cpp, ggml) and Hugging Face community.