File size: 6,049 Bytes

248dc1e

---
base_model:
- openbmb/MiniCPM5-1B
pipeline_tag: text-generation
tags:
- quantized
- nvfp4
- autoround
- low-bit-open-llm-leaderboard
---

# MiniCPM5-1B-AutoRound-NVFP4-RTN

## Model Details

This model is a NVFP4 (NVIDIA FP4) quantization of [openbmb/MiniCPM5-1B](https://huggingface.co/openbmb/MiniCPM5-1B) generated by [AutoRound](https://github.com/intel/auto-round). Please follow the license of the original model.

## Quantization Details

| Attribute | Value |
|-----------|-------|
| Base Model | [openbmb/MiniCPM5-1B](https://huggingface.co/openbmb/MiniCPM5-1B) |
| Quantization Tool | [AutoRound](https://github.com/intel/auto-round) |
| Quantization Scheme | NVFP4 |
| Original Size | 1089 MB |
| Quantized Size | 1363 MB |

## Evaluation Results

| Task | Accuracy |
|------|----------|
| hellaswag | 0.3691 |
| mmlu | 0.4870 |
| mmlu_abstract_algebra | 0.3000 |
| mmlu_anatomy | 0.5407 |
| mmlu_astronomy | 0.5395 |
| mmlu_business_ethics | 0.4500 |
| mmlu_clinical_knowledge | 0.5434 |
| mmlu_college_biology | 0.5486 |
| mmlu_college_chemistry | 0.3800 |
| mmlu_college_computer_science | 0.4600 |
| mmlu_college_mathematics | 0.3800 |
| mmlu_college_medicine | 0.4855 |
| mmlu_college_physics | 0.3235 |
| mmlu_computer_security | 0.5700 |
| mmlu_conceptual_physics | 0.4000 |
| mmlu_econometrics | 0.2982 |
| mmlu_electrical_engineering | 0.5517 |
| mmlu_elementary_mathematics | 0.3519 |
| mmlu_formal_logic | 0.3413 |
| mmlu_global_facts | 0.2100 |
| mmlu_high_school_biology | 0.5806 |
| mmlu_high_school_chemistry | 0.4236 |
| mmlu_high_school_computer_science | 0.4500 |
| mmlu_high_school_european_history | 0.6061 |
| mmlu_high_school_geography | 0.5859 |
| mmlu_high_school_government_and_politics | 0.6321 |
| mmlu_high_school_macroeconomics | 0.4692 |
| mmlu_high_school_mathematics | 0.2926 |
| mmlu_high_school_microeconomics | 0.5042 |
| mmlu_high_school_physics | 0.2649 |
| mmlu_high_school_psychology | 0.6624 |
| mmlu_high_school_statistics | 0.3426 |
| mmlu_high_school_us_history | 0.5588 |
| mmlu_high_school_world_history | 0.6160 |
| mmlu_human_aging | 0.4888 |
| mmlu_human_sexuality | 0.6260 |
| mmlu_humanities | 0.4389 |
| mmlu_international_law | 0.7107 |
| mmlu_jurisprudence | 0.5926 |
| mmlu_logical_fallacies | 0.5828 |
| mmlu_machine_learning | 0.3839 |
| mmlu_management | 0.6408 |
| mmlu_marketing | 0.7521 |
| mmlu_medical_genetics | 0.6600 |
| mmlu_miscellaneous | 0.6564 |
| mmlu_moral_disputes | 0.5087 |
| mmlu_moral_scenarios | 0.2380 |
| mmlu_nutrition | 0.6307 |
| mmlu_other | 0.5555 |
| mmlu_philosophy | 0.5466 |
| mmlu_prehistory | 0.5278 |
| mmlu_professional_accounting | 0.3723 |
| mmlu_professional_law | 0.3677 |
| mmlu_professional_medicine | 0.4669 |
| mmlu_professional_psychology | 0.4935 |
| mmlu_public_relations | 0.5000 |
| mmlu_security_studies | 0.5714 |
| mmlu_social_sciences | 0.5583 |
| mmlu_sociology | 0.6766 |
| mmlu_stem | 0.4218 |
| mmlu_us_foreign_policy | 0.6700 |
| mmlu_virology | 0.4578 |
| mmlu_world_religions | 0.7193 |
| piqa | 0.6670 |

## How to Use

### HF Usage

**Step 1: Install [AutoRound](https://github.com/intel/auto-round)**

```bash
pip install auto-round
```

**Step 2: Load and run the quantized model**

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "MiniCPM5-1B-AutoRound-NVFP4-RTN"

# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")

# prepare the model input
prompt = "Write a quick sort algorithm."
messages = [{"role": "user", "content": prompt}]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

# conduct text completion
generated_ids = model.generate(**model_inputs, max_new_tokens=512)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]) :].tolist()

content = tokenizer.decode(output_ids, skip_special_tokens=True)
print("content:", content)
```

### VLLM Usage

```bash
vllm serve MiniCPM5-1B-AutoRound-NVFP4-RTN \
    --trust-remote-code \
    --dtype bfloat16 \
    --tensor_parallel_size 1
```

If you encounter any issues, feel free to open an issue on the [AutoRound GitHub repo](https://github.com/intel/auto-round/issues) or provide feedback on the [Low-Bit Open LLM Leaderboard](https://huggingface.co/spaces/Intel/low_bit_open_llm_leaderboard).

## Ethical Considerations and Limitations

The model can produce factually incorrect output, and should not be relied on to produce factually accurate information. Because of the limitations of the pretrained model and the finetuning datasets, it is possible that this model could generate lewd, biased or otherwise offensive outputs.
Therefore, before deploying any applications of the model, developers should perform safety testing.

## Caveats and Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model.
Here are a couple of useful links to learn more about Intel's AI software:

- [Intel Neural Compressor](https://github.com/intel/neural-compressor)
- [AutoRound](https://github.com/intel/auto-round)

## Disclaimer

The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. Please consult an attorney before using this model for commercial purposes.

## Cite

```
@article{cheng2023optimize,
  title={Optimize weight rounding via signed gradient descent for the quantization of llms},
  author={Cheng, Wenhua and Zhang, Weiwei and Shen, Haihao and Cai, Yiyang and He, Xin and Lv, Kaokao and Liu, Yi},
  journal={arXiv preprint arXiv:2309.05516},
  year={2023}
}
```

[arxiv](https://arxiv.org/abs/2309.05516) [github](https://github.com/intel/auto-round)

---

*This model is part of the [Intel Low-Bit Open LLM Leaderboard](https://huggingface.co/spaces/Intel/low_bit_open_llm_leaderboard) initiative.*