---
license: other
language:
- vi
model_name: PhoGPT 7B5 Instruct
inference: false
model_creator: VinAI Research
model_link: https://huggingface.co/vinai/PhoGPT-7B5-Instruct
model_type: mpt
pipeline_tag: text-generation
quantized_by: nguyenviet
base_model: vinai/PhoGPT-7B5-Instruct
---

# PhoGPT-7B5-Instruct.GGUF

GGUF format files of the model [vinai/PhoGPT-7B5-Instruct](https://huggingface.co/vinai/PhoGPT-7B5-Instruct).

## Model Details

For detailed information about the original model, please refer to [phoGPT's repository](https://github.com/VinAIResearch/PhoGPT).

## Uses

Select and download the quantization version that fits the needs.

## License

PhoGPT is licensed under the [PhoGPT Community License](https://github.com/VinAIResearch/PhoGPT/blob/main/LICENSE), Copyright (c) VinAI. All Rights Reserved.

## Provided files

| Name | Quant method | Size | Use case |
| ---- | ---- | ---- | ----- |
| [PhoGPT-7B5-Instruct-q2_k.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q2_k.gguf) | Q2_K | 3.8 GB | smallest, significant quality loss - not recommended for most purposes |
| [PhoGPT-7B5-Instruct-q3_k_s.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q3_k_s.gguf) | Q3_K_S | 4.07 GB | very small, high quality loss |
| [PhoGPT-7B5-Instruct-q3_k_m.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q3_k_m.gguf) | Q3_K_M | 4.66 GB | very small, high quality loss |
| [PhoGPT-7B5-Instruct-q3_k_l.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q3_k_l.gguf) | Q3_K_L | 4.98 GB | small, substantial quality loss |
| [PhoGPT-7B5-Instruct-q4_0.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q4_0.gguf) | Q4_0 | 5.06 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
| [PhoGPT-7B5-Instruct-q4_k_s.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q4_k_s.gguf) | Q4_K_S | 5.1 GB | small, greater quality loss |
| [PhoGPT-7B5-Instruct-q4_k_m.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q4_k_m.gguf) | Q4_K_M | 5.54 GB | medium, balanced quality - recommended |
| [PhoGPT-7B5-Instruct-q4_1.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q4_1.gguf) | Q4_1 | 5.53 GB | legacy; higher accuracy than Q4_0 but not as high as Q5_0, however has quicker inference than Q5 models.
| [PhoGPT-7B5-Instruct-q5_0.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q5_0.gguf) | Q5_0 | 6 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
| [PhoGPT-7B5-Instruct-q5_k_s.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q5_k_s.gguf) | Q5_K_S | 6 GB | large, low quality loss - recommended |
| [PhoGPT-7B5-Instruct-q5_k_m.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q5_k_m.gguf) | Q5_K_M | 6.35 GB | large, very low quality loss - recommended |
| [PhoGPT-7B5-Instruct-q5_1.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q5_1.gguf) | Q5_1 | 6.46 GB | legacy; even higher accuracy, resource usage and slower inference.
| [PhoGPT-7B5-Instruct-q6_k.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q6_k.gguf) | Q6_K | 6.99 GB | very large, extremely low quality loss |
| [PhoGPT-7B5-Instruct-q8_0.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-q8_0.gguf) | Q8_0 | 9.05 GB | almost indistinguishable from float16. High resource use and slow, not recommended for most users |
| [PhoGPT-7B5-Instruct-f16.gguf](https://huggingface.co/nguyenviet/PhoGPT-7B5-Instruct-GGUF/blob/main/PhoGPT-7B5-Instruct-f16.gguf) | float16 | 17 GB | very large, extremely low quality loss - not recommended |