Search papers, labs, and topics across Lattice.
This paper introduces JoyNexus, a service-oriented framework designed for multi-tenant post-training of Vision-Language-Action (VLA) models, addressing inefficiencies in existing compute services that allocate exclusive resources to single tenants. By decoupling training, inference, and environment services and implementing group batching for heterogeneous data, JoyNexus allows multiple tenants to concurrently submit workloads while maintaining isolation of their components. The evaluation demonstrates that JoyNexus significantly reduces aggregate GPU time and enhances service utilization compared to traditional single-tenant execution models.
JoyNexus slashes GPU time and boosts service efficiency by enabling concurrent multi-tenant training with isolated workloads and shared resources.
The post-training of Vision-Language-Action (VLA) models is essential due to the diversity of simulators, robot embodiments, and task objectives. Existing compute services, whether offered as direct accelerator rental or batch-workload submission, typically allocate an exclusive set of GPU and CPU resources to a single tenant. While this paradigm maximizes client flexibility, it burdens users with infrastructure adaptation, and the fixed card-hour accounting model renders short or bursty workloads both expensive for tenants and inefficient for the service provider. To address these challenges, we present JoyNexus, a unified service for multi-tenant VLA supervised fine-tuning, reinforcement learning, and evaluation. JoyNexus decouples the Training Model Service, Inference Model Service, and Environment Service, each accessed through APIs and backed by resident shared base models with tenant-specific slots. Tenants can directly invoke high-level semantic APIs for training, rollout, and evaluation, or compose custom algorithms using lower-level APIs and their assigned endpoints. Multiple tenants submit workloads concurrently; their action modules, optimizers, rollout records, and policy versions remain isolated, and the service is scheduled by the global Training Queue and Inference Queue. To further improve multi-tenant training efficiency, JoyNexus introduces group batching for heterogeneous VLA data schemas that share a compatible model-facing prefix, enabling a single shared backbone forward pass over grouped samples. Finally, we evaluate JoyNexus through workload simulation and a group-batching pipeline in a realistic embodied scenario. Results show that, compared with isolated single-tenant execution, JoyNexus reduces aggregate GPU time and improves service utilization via cross-tenant scheduling on shared resources.
