xmadai/Llama-3.1-Nemotron-70B-Instruct-xMADai-INT4

This model has been xMADified!

This repository contains nvidia/Llama-3.1-Nemotron-70B-Instruct-HF quantized from 16-bit floats to 4-bit integers, using xMAD.ai proprietary technology.

Why should I use this model?

1. Accuracy: This xMADified model is the best quantized version of the nvidia/Llama-3.1-Nemotron-70B-Instruct-HF model (40 GB only). See Table 1 below for model quality benchmarks.

2. Memory-efficiency: The full-precision model is around 140 GB, while this xMADified model is under 40 GB, making it feasible to run on a single 48 GB GPU.

3. Fine-tuning: These models are fine-tunable over the same reduced (a single 48 GB GPU) hardware in mere 3-clicks. Watch our product demo here

Table 1: xMAD vs. Unsloth

Model	Arc Challenge	Arc Easy	LAMBADA Standard	LAMBADA OpenAI	MMLU STEM	MMLU Humanities	MMLU Social Sciences	MMLU Other
xmadai/Llama-3.1-Nemotron-70B-Instruct-xMADai-INT4 (this model)	63.05	86.36	71.96	75.82	75.55	80.62	87.42	83.71
unsloth/Llama-3.1-Nemotron-70B-Instruct-bnb-4bit	60.32	85.35	71.01	74.64	75.1	80.06	87.33	83.71

How to Run Model

Loading the model checkpoint of this xMADified model requires around 40 GB of VRAM. Hence it can be efficiently run on a single 48 GB GPU.

Package prerequisites:

1. Run the following *commands to install the required packages.

pip install torch==2.4.0  # Run following if you have CUDA version 11.8: pip install torch==2.4.0 --index-url https://download.pytorch.org/whl/cu118
pip install transformers accelerate optimum
pip install -vvv --no-build-isolation "git+https://github.com/PanQiWei/[email protected]"

Sample Inference Code

from transformers import AutoTokenizer
from auto_gptq import AutoGPTQForCausalLM
model_id = "xmadai/Llama-3.1-Nemotron-70B-Instruct-xMADai-INT4"
prompt = [
    {"role": "system", "content": "You are a helpful assistant, that responds as a pirate."},
    {"role": "user", "content": "What's Deep Learning?"},
]
tokenizer = AutoTokenizer.from_pretrained(model_id, use_fast=False)
inputs = tokenizer.apply_chat_template(
    prompt,
    tokenize=True,
    add_generation_prompt=True,
    return_tensors="pt",
    return_dict=True,
).to("cuda")
model = AutoGPTQForCausalLM.from_quantized(
    model_id,
    device_map='auto',
    trust_remote_code=True,
)
outputs = model.generate(**inputs, do_sample=True, max_new_tokens=1024)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))

Citation

If you found this model useful, please cite our research paper.

@article{zhang2024leanquant,
  title={LeanQuant: Accurate and Scalable Large Language Model Quantization with Loss-error-aware Grid},
  author={Zhang, Tianyi and Shrivastava, Anshumali},
  journal={arXiv preprint arXiv:2407.10032},
  year={2024},
  url={https://arxiv.org/abs/2407.10032},
}

Contact Us

For additional xMADified models, access to fine-tuning, and general questions, please contact us at [email protected] and join our waiting list.