Scaling Multimodal Search and Recommendation with Small Language Models via Upside-Down Reinforcement Learning
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866914177973485568 |
|---|---|
| author | Lin, Yu-Chen Sharma, Sanat Manikandan, Hari Kumar, Jayant King, Tracy Holloway Zheng, Jing |
| author_facet | Lin, Yu-Chen Sharma, Sanat Manikandan, Hari Kumar, Jayant King, Tracy Holloway Zheng, Jing |
| contents | In this work, we investigate how small language models (SLMs) can be scaled to support multimodal search and recommendation use cases while remaining efficient enough for real-time, resource-constrained deployments. We present a framework that combines upside-down reinforcement learning with synthetic data distillation from a large language model (Llama-3) to train a 100M-parameter GPT-2 model for multitask prompt generation. Despite being up to 80 times smaller than state-of-the-art large language models (LLMs), our SLM achieves relevance and diversity scores within 6% of competitive baselines such as Llama-3 8B, Qwen3 8B, and Ministral 8B. These results demonstrate that SLMs can effectively handle multimodal search and recommendation tasks, while dramatically reducing inference latency and memory overhead. Our study highlights the potential of lightweight models as practical engines for scalable multimodal discovery, bridging the gap between cutting-edge research and real-world multimodal applications such as media recommendations and creative content generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_09854 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Scaling Multimodal Search and Recommendation with Small Language Models via Upside-Down Reinforcement Learning Lin, Yu-Chen Sharma, Sanat Manikandan, Hari Kumar, Jayant King, Tracy Holloway Zheng, Jing Computation and Language Artificial Intelligence Machine Learning In this work, we investigate how small language models (SLMs) can be scaled to support multimodal search and recommendation use cases while remaining efficient enough for real-time, resource-constrained deployments. We present a framework that combines upside-down reinforcement learning with synthetic data distillation from a large language model (Llama-3) to train a 100M-parameter GPT-2 model for multitask prompt generation. Despite being up to 80 times smaller than state-of-the-art large language models (LLMs), our SLM achieves relevance and diversity scores within 6% of competitive baselines such as Llama-3 8B, Qwen3 8B, and Ministral 8B. These results demonstrate that SLMs can effectively handle multimodal search and recommendation tasks, while dramatically reducing inference latency and memory overhead. Our study highlights the potential of lightweight models as practical engines for scalable multimodal discovery, bridging the gap between cutting-edge research and real-world multimodal applications such as media recommendations and creative content generation. |
| title | Scaling Multimodal Search and Recommendation with Small Language Models via Upside-Down Reinforcement Learning |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2502.09854 |