LLaMA-E

Model Introduction

LLaMA-E is meticulously crafted for e-commerce authoring tasks, incorporating specialized features to excel in generating product descriptions, advertisements, and other related content, as outlined in https://arxiv.org/abs/2308.04913#/.

Usage

Inference using Huggingface transformers on NVIDIA GPUs. Requirements tested on python 3.10:

python inference.py --load_8bit --base_model 'meta-llama/Llama-2-7b-hf' --lora_weights 'DSMI/LLaMA-E'

***Instruction: Where can I buy the handmade jewellery?
***Response: Etsy, NOVICA, UncommonGoods, World Market, and Global Crafts offer unique handmade jewelry from artisans worldwide.

***Instruction: Create an attractive advertisement for the Christmas sale of the following product.
***Input: Custom Photo Music Plaque,Personalized Photo Frame,Album Cover Song Plaque,Music Photo Name Night Lamp,Photo and Music Gift, Music Prints
***Response: Personalize your music space with a Custom Photo Music Plaque. Add your favorite photo and song lyrics to create a unique gift for music lovers.

Citation

If you find our work helpful, please consider citing the following papers.

@inproceedings{shi-etal-2025-llama,
    title = "{LL}a{MA}-{E}: Empowering {E}-commerce Authoring with Object-Interleaved Instruction Following",
    author = "Shi, Kaize  and
      Sun, Xueyao  and
      Wang, Dingxian  and
      Fu, Yinlin  and
      Xu, Guandong  and
      Li, Qing",
    editor = "Rambow, Owen  and
      Wanner, Leo  and
      Apidianaki, Marianna  and
      Al-Khalifa, Hend  and
      Eugenio, Barbara Di  and
      Schockaert, Steven",
    booktitle = "Proceedings of the 31st International Conference on Computational Linguistics",
    month = jan,
    year = "2025",
    address = "Abu Dhabi, UAE",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.coling-main.58/",
    pages = "870--885",
    abstract = "E-commerce authoring entails creating engaging, diverse, and targeted content to enhance preference elicitation and retrieval experience. While Large Language Models (LLMs) have revolutionized content generation, they often fall short in e-commerce applications due to their limited memorization of domain-specific features. This paper proposes LLaMA-E, the unified e-commerce authoring models that address the contextual preferences of customers, sellers, and platforms, the essential objects in e-commerce operation. We design the instruction set derived from tasks of ads generation, query-enhanced product title rewriting, product classification, purchase intent speculation, and general e-commerce Q{\&}A. The instruction formulation ensures the interleaved cover of the presented and required object features, allowing the alignment of base models to parameterize e-commerce knowledge comprehensively. The proposed LLaMA-E models achieve state-of-the-art evaluation performance and exhibit the advantage in zero-shot practical applications. To our knowledge, this is the first LLM tailored to empower authoring applications with comprehensive scenario understanding by integrating features focused on participated objects."
}

License

The model released here is under the Llama-2 LICENSE to ensure more flexible accessibility; please adhere to the corresponding licence.

Acknowledgements

Our code for the inference is based on the tloen.