GrocLM: LLM enhances e-commerce grocery recommendations

Researchers have developed GrocLM, a large language model specifically designed for category-level recommendations in e-commerce grocery shopping. This model utilizes a two-stage LoRA-based training approach to capture cyclical purchasing behaviors and rebuying signals more effectively than traditional methods. GrocLM also incorporates a trie-based constrained decoding mechanism to ensure accurate and controllable outputs within a predefined category space. In production testing for restocking tasks, GrocLM demonstrated a 7.5% relative improvement in cart-adds per impression, showcasing its practical effectiveness and efficiency. AI

IMPACT This model could improve online grocery shopping experiences by providing more relevant category recommendations, potentially increasing sales for e-commerce platforms.

RANK_REASON The cluster describes a research paper detailing a new model and its performance on a specific task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• CORE Recommender
• DagsHub
• e-commerce
• GrocLM
• Hugging Face
• large-language models
• Lora

• model release
• product
• paper

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →