gemma-4-A4B-98e-v6-coder-it-AutoRound-MXFP4-RTN

Model Details

This model is a MXFP4 (Microscaling FP4) quantization of ManniX-ITA/gemma-4-A4B-98e-v6-coder-it generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	ManniX-ITA/gemma-4-A4B-98e-v6-coder-it
Quantization Tool	AutoRound
Quantization Scheme	MXFP4
Original Size	4619 MB
Quantized Size	13756 MB

Evaluation Results

Task	Accuracy
hellaswag	0.3211
mmlu	0.3358
mmlu_abstract_algebra	0.3300
mmlu_anatomy	0.2963
mmlu_astronomy	0.3289
mmlu_business_ethics	0.3400
mmlu_clinical_knowledge	0.3547
mmlu_college_biology	0.4514
mmlu_college_chemistry	0.2900
mmlu_college_computer_science	0.4600
mmlu_college_mathematics	0.3200
mmlu_college_medicine	0.3064
mmlu_college_physics	0.2353
mmlu_computer_security	0.4000
mmlu_conceptual_physics	0.3191
mmlu_econometrics	0.3246
mmlu_electrical_engineering	0.2621
mmlu_elementary_mathematics	0.3148
mmlu_formal_logic	0.3730
mmlu_global_facts	0.2100
mmlu_high_school_biology	0.4226
mmlu_high_school_chemistry	0.3251
mmlu_high_school_computer_science	0.4300
mmlu_high_school_european_history	0.4303
mmlu_high_school_geography	0.2828
mmlu_high_school_government_and_politics	0.2902
mmlu_high_school_macroeconomics	0.3077
mmlu_high_school_mathematics	0.2481
mmlu_high_school_microeconomics	0.3824
mmlu_high_school_physics	0.3377
mmlu_high_school_psychology	0.3761
mmlu_high_school_statistics	0.3796
mmlu_high_school_us_history	0.4804
mmlu_high_school_world_history	0.5021
mmlu_human_aging	0.3094
mmlu_human_sexuality	0.3359
mmlu_humanities	0.3360
mmlu_international_law	0.4050
mmlu_jurisprudence	0.2685
mmlu_logical_fallacies	0.3497
mmlu_machine_learning	0.3304
mmlu_management	0.2816
mmlu_marketing	0.4316
mmlu_medical_genetics	0.3600
mmlu_miscellaneous	0.2848
mmlu_moral_disputes	0.3324
mmlu_moral_scenarios	0.2994
mmlu_nutrition	0.3464
mmlu_other	0.3183
mmlu_philosophy	0.3473
mmlu_prehistory	0.3086
mmlu_professional_accounting	0.2766
mmlu_professional_law	0.2999
mmlu_professional_medicine	0.3272
mmlu_professional_psychology	0.3399
mmlu_public_relations	0.3545
mmlu_security_studies	0.4082
mmlu_social_sciences	0.3503
mmlu_sociology	0.3731
mmlu_stem	0.3387
mmlu_us_foreign_policy	0.4700
mmlu_virology	0.3373
mmlu_world_religions	0.3509
piqa	0.5288

How to Use

HF Usage

Step 1: Install AutoRound

pip install auto-round

Step 2: Load and run the quantized model

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "gemma-4-A4B-98e-v6-coder-it-AutoRound-MXFP4-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

vllm serve gemma-4-A4B-98e-v6-coder-it-AutoRound-MXFP4-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 or provide feedback on the 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:

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 github

This model is part of the Intel Low-Bit Open LLM Leaderboard initiative.