Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.24273 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908995299573760 |
|---|---|
| author | Sajid, Md. Ashiq Ul Islam Mahmood, Mohammad Sakib Hasan, Md. Tareq Rahim, Md Abdur Ara, Rafat Hossain, Md. Arafat |
| author_facet | Sajid, Md. Ashiq Ul Islam Mahmood, Mohammad Sakib Hasan, Md. Tareq Rahim, Md Abdur Ara, Rafat Hossain, Md. Arafat |
| contents | The deployment of intelligent reinforcement learning (RL) agents on resource-constrained edge devices remains a fundamental challenge due to the substantial memory, computational, and energy requirements of modern deep learning systems. While large language models (LLMs) have emerged as powerful architectures for decision-making agents, their multi-billion parameter scale confines them to cloud-based deployment, raising concerns about latency, privacy, and connectivity dependence.
We introduce BitRL, a framework for building RL agents using 1-bit quantized language models that enables practical on-device learning and inference under severe resource constraints. Leveraging the BitNet b1.58 architecture with ternary weights (-1, 0, +1) and an optimized inference stack, BitRL achieves 10-16x memory reduction and 3-5x energy efficiency improvements over full-precision baselines while maintaining 85-98 percent of task performance across benchmarks.
We provide theoretical analysis of quantization as structured parameter perturbation, derive convergence bounds for quantized policy gradients under frozen-backbone architectures, and identify the exploration-stability trade-off in extreme quantization. Our framework systematically integrates 1-bit quantized language models with reinforcement learning for edge deployment and demonstrates effectiveness on commodity hardware. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_24273 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | BitRL: Reinforcement Learning with 1-bit Quantized Language Models for Resource-Constrained Edge Deployment Sajid, Md. Ashiq Ul Islam Mahmood, Mohammad Sakib Hasan, Md. Tareq Rahim, Md Abdur Ara, Rafat Hossain, Md. Arafat Machine Learning The deployment of intelligent reinforcement learning (RL) agents on resource-constrained edge devices remains a fundamental challenge due to the substantial memory, computational, and energy requirements of modern deep learning systems. While large language models (LLMs) have emerged as powerful architectures for decision-making agents, their multi-billion parameter scale confines them to cloud-based deployment, raising concerns about latency, privacy, and connectivity dependence. We introduce BitRL, a framework for building RL agents using 1-bit quantized language models that enables practical on-device learning and inference under severe resource constraints. Leveraging the BitNet b1.58 architecture with ternary weights (-1, 0, +1) and an optimized inference stack, BitRL achieves 10-16x memory reduction and 3-5x energy efficiency improvements over full-precision baselines while maintaining 85-98 percent of task performance across benchmarks. We provide theoretical analysis of quantization as structured parameter perturbation, derive convergence bounds for quantized policy gradients under frozen-backbone architectures, and identify the exploration-stability trade-off in extreme quantization. Our framework systematically integrates 1-bit quantized language models with reinforcement learning for edge deployment and demonstrates effectiveness on commodity hardware. |
| title | BitRL: Reinforcement Learning with 1-bit Quantized Language Models for Resource-Constrained Edge Deployment |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2604.24273 |