Reproduce “FlashX: Scaling Robotic Experience Reuse for Embodied Agents with CBA-Based NAND Flash” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuseon Choi, Sanghyeon Park, Eunjong Lee, Hanseo Lee, Sangjin Kim
FlashX co-designs NAND storage and search for experience reuse in robots, placing retrieval keys in faster pSLC and latent values in denser TLC, with interleaved in-flash engines. Retrieved latents warm-start action denoising and adapt to observations; low-similarity matches trigger full inference. The system reports up to 12.24-point success gains and 2.57× search throughput, based on modeled hardware, not a fabricated device.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.