Research map / Reinforcement learning / Visual and representation RL
DyGRO-VLA: Cross-Task Scaling of Vision–Language–Action Models via Dynamic Grouped Residual Optimization
Sixu Lin, Yunpeng Qing, Litao Liu, Ming Zhou, Ruixing Jin, Xiaoyi Fan, Guiliang Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add reinforcement fine-tuning (online MoRR) to offline base | DyGRO-VLA · LIBERO Avg | success rate (%): 92.7 -> 97.1 | Improved |
| add reinforcement fine-tuning to offline base | DyGRO-VLA · LIBERO-Long | success rate (%): 85.2 -> 95.0 | Improved |
| increase mixture experts from 1 to 8 (ablation) | DyGRO-VLA MoRR · LIBERO Avg | success rate (%): 93.4 -> 97.1 | Improved |
| increase experts from 4 to 8 (ablation) | DyGRO-VLA MoRR · LIBERO Avg | success rate (%): 97.0 -> 97.1 | No clear effect |
| add information bottleneck objective (ablation) | DyGRO-VLA · LIBERO Avg | success rate (%): 96.8 -> 97.1 | Improved |
| add InfoNCE contrastive objective (ablation) | DyGRO-VLA · LIBERO Avg | success rate (%): 94.3 -> 97.1 | Improved |
| add InfoNCE contrastive objective (ablation) | DyGRO-VLA · LIBERO-Long | success rate (%): 90.4 -> 95.0 | Improved |
| add difficulty-aware sampling (ablation) | DyGRO-VLA · LIBERO Avg | success rate (%): 96.4 -> 97.1 | Improved |
| replace shared MLP residual head with MoRR dynamic grouping (ablation) | DyGRO-VLA · Gradient cosine similarity (Open/Close tasks) | gradient similarity | Improved |
| update VLM backbone with LoRA in online RL (ablation) | VLM (LoRA) + Action Expert · LIBERO online training | success rate | Hurt |
| use standard residual RL instead of MoRR (ablation) | Standard Residual RL · LIBERO online training | success rate | Hurt |
| DyGRO-VLA vs OpenVLA-oft RFT on Sim2Real | DyGRO-VLA · RoboTwin2 Avg (Real) | success rate (%): 55.0 -> 57.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SimpleVLA-RL — SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning (ICLR 2026) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- VLA-R1: Enhancing Reasoning in Vision-Language-Action Models (ECCV 2026 · similarity 0.92)
- What Can RL Bring to VLA Generalization? An Empirical Study (NeurIPS 2025 · similarity 0.92)
- NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Models (NeurIPS 2026 · similarity 0.91)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.91)
- Beyond Data Scaling: Representation-Centric Pre-training for Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
- VLA-REPLICA: A Low-Cost, Reproducible Benchmark for Real-World Evaluation of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
