iiinova — where your digital self keeps evolving

Krishna Acharya

@kvachai.bsky.social

13 followers 20 following 11 posts

Ph.D candidate @Georgia Tech | Recommender systems, Algorithmic Fairness, Differential privacy. krishnacharya.github.io

Posts Replies Media

Krishna Acharya @kvachai.bsky.social · 09/06/2025

8/8 🗓️ I’ll be at the KDD Workshop on Online and Adaptive Recommender Systems (OARS) — happy to chat about this work, online and in person in Toronto! #GLoSS #KDD2025 #OARS #LLM #RecommenderSystems #SemanticSearch #DenseRetrieval #LoRA #LLaMA3

010

Krishna Acharya @kvachai.bsky.social · 09/06/2025

7/8 In addition, user segment-wise evaluation shows: - Strong gains for cold-start users in Toys and Sports - Benefits from longer user histories in Beauty This highlights GLoSS’s robustness across interaction lengths. #ColdStart #Personalization

100

Krishna Acharya @kvachai.bsky.social · 09/06/2025

6/8 📈 Results on Amazon Beauty, Toys, and Sports datasets, GLoSS improves : Recall@5 by +33.3%, +52.8%, +15.2% - NDCG@5 by +30.0%, +42.6%, +16.1% over ID-based baselines. GLoSS also outperforms LLM-based models(P5, GPT4Rec, LlamaRec, E4SRec) with Recall@5 gains of +4.3%, +22.8%, +29.5% respectively.

100

Krishna Acharya @kvachai.bsky.social · 09/06/2025

5/8 For query generation, we fine-tune 4-bit quantized LLaMA-3 models (1B, 3B, 8B) using LoRA— enabling efficient training on a single RTX A5000 using the Unsloth AI library. For dense retrieval, we use e5-small-v2 as the text encoder. #LoRA #LLaMA3 #Unsloth

100

Krishna Acharya @kvachai.bsky.social · 09/06/2025

4/8 Prior LLM-based recommenders often rely on lexical search methods like BM25. GLoSS instead uses dense retrieval, going beyond frequency-based token overlap to capture deeper semantic relevance.

100

Krishna Acharya @kvachai.bsky.social · 09/06/2025

3/8 Classic ID-based approaches like SASRec, BERT4Rec, and SemanticID based models like TIGER are effective— but usually require retraining when new items are added and struggle to generalize beyond patterns seen in training data, especially without rich metadata.

100

Krishna Acharya @kvachai.bsky.social · 09/06/2025

2/8 GLoSS is a generative recommendation framework that integrates LLMs with semantic search (aka dense retrieval) for sequential recommendation. #LLM #RecommenderSystems #DenseRetrieval

110

Krishna Acharya @kvachai.bsky.social · 09/06/2025

1/8 Happy to share that our paper GLoSS: Generative Language Models with Semantic Search for Sequential Recommendation is accepted at the KDD OARS workshop! 🎉 Paper, code: github.com/krishnachary... This is joint work with my wonderful collaborators @apetrov.bsky.social and @jubaz.bsky.social !

github.com

GitHub - krishnacharya/GLoSS: GLoSS: Generative Language Models with Semantic Search for Sequential Recommendation

GLoSS: Generative Language Models with Semantic Search for Sequential Recommendation - krishnacharya/GLoSS

111

Krishna Acharya @kvachai.bsky.social · 24/04/2025

3/3 Among these baselines, a classic retrieval approach (using BM25) based on the text of the last item performs the best. I also explore how often-overlooked steps, like failing to deduplicate exact user-item interactions, can lead to significant inflation in metrics.

000

Krishna Acharya @kvachai.bsky.social · 24/04/2025

2/3 In this post, I dive into different model types—from ID-based to fully metadata-based models, key preprocessing steps, the leave-one-item-out split, evaluation metrics, and four baselines that any trained recommender should aim to beat.

100

Krishna Acharya @kvachai.bsky.social · 24/04/2025

Happy to share that I’ve started writing! Check out my first post on generative recommendation here: krishnacharya.github.io/posts/genera...

110