VLA language backbones show high redundancy on manipulation benchmarks, with half the LLM blocks removable and even two blocks sufficient to recover baseline performance after fine-tuning, unlike vision and action pathways.
super hub Mixed citations
Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success
Mixed citation behavior. Most common role is background (64%).
165
Pith papers citing it
Background
64% of classified citations
abstract
Recent vision-language-action models (VLAs) build upon pretrained vision-language models and leverage diverse robot datasets to demonstrate strong task execution, language following ability, and semantic generalization. Despite these successes, VLAs struggle with novel robot setups and require fine-tuning to achieve good performance, yet how to most effectively fine-tune them is unclear given many possible strategies. In this work, we study key VLA adaptation design choices such as different action decoding schemes, action representations, and learning objectives for fine-tuning, using OpenVLA as our representative base model. Our empirical analysis informs an Optimized Fine-Tuning (OFT) recipe that integrates parallel decoding, action chunking, a continuous action representation, and a simple L1 regression-based learning objective to altogether improve inference efficiency, policy performance, and flexibility in the model's input-output specifications. We propose OpenVLA-OFT, an instantiation of this recipe, which sets a new state of the art on the LIBERO simulation benchmark, significantly boosting OpenVLA's average success rate across four task suites from 76.5% to 97.1% while increasing action generation throughput by 26$\times$. In real-world evaluations, our fine-tuning recipe enables OpenVLA to successfully execute dexterous, high-frequency control tasks on a bimanual ALOHA robot and outperform other VLAs ($\pi_0$ and RDT-1B) fine-tuned using their default recipes, as well as strong imitation learning policies trained from scratch (Diffusion Policy and ACT) by up to 15% (absolute) in average success rate. We release code for OFT and pretrained model checkpoints at https://openvla-oft.github.io/.
hub tools
citation-role summary
background 34
baseline 10
method 6
other 3
citation-polarity summary
claims ledger
- abstract Recent vision-language-action models (VLAs) build upon pretrained vision-language models and leverage diverse robot datasets to demonstrate strong task execution, language following ability, and semantic generalization. Despite these successes, VLAs struggle with novel robot setups and require fine-tuning to achieve good performance, yet how to most effectively fine-tune them is unclear given many possible strategies. In this work, we study key VLA adaptation design choices such as different action decoding schemes, action representations, and learning objectives for fine-tuning, using OpenVLA
authors
co-cited works
representative citing papers
LIBERO-Safety supplies a scalable benchmark, data-generation pipeline, and 19,664-demonstration dataset that exposes a generalization-safety tension in current VLA models where diverse training improves collision avoidance but task success stays limited by trajectory quality and semantic understandi
BOKBO is the first conformal abstention method for K-sample VLA policies that supplies finite-sample distribution-free guarantees on executed violation rates, with global and Mondrian per-task variants.
JOPAT jointly models pixels, point tracks, and actions in a diffusion transformer and reports gains over pixel-only baselines on long-horizon robot tasks with occlusion and off-screen motion.
Pion modifies Muon's Newton-Schulz iterations into a controllable high-pass filter that anchors dominant singular values at 1 while suppressing noisy tails, outperforming Muon and AdamW in VLA and RLVR regimes.
PCM uses success-failure action variance to probabilistically select and mask chunks for gradient updates in GRPO, matching standard success rates with 2.38x wall-clock speedup and 60% lower memory on LIBERO benchmarks.
DSSP is a history-conditioned diffusion state space policy that uses SSMs to encode full observation streams with an auxiliary dynamics objective and hierarchical fusion, achieving SOTA results with reduced model size in robot manipulation.
RotVLA models latent actions as continuous SO(n) rotations with triplet-frame supervision and flow-matching to reach 98.2% success on LIBERO and 89.6%/88.5% on RoboTwin2.0 using a 1.7B-parameter model.
GridS is a plug-and-play differentiable module for geometry-aware visual token resampling in VLA models that achieves under 10% token retention and 76% FLOPs reduction with no success-rate loss.
MCF-Proto adds a motion-centric local action frame and prototype parameterization to VLA models, inducing emergent geometric structure and improved robustness from standard demonstrations alone.
Capability vectors extracted from parameter differences between standard and auxiliary-finetuned VLA models can be merged into pretrained weights to match auxiliary-training performance while reducing computational overhead during adaptation.
VEGA improves spatial reasoning in VLA models for robotics by aligning visual encoder features with 3D-supervised DINOv2 representations via a temporary projector and cosine similarity loss.
GuardVLA embeds a stealthy backdoor watermark in VLAs via secret messages in visual data and uses a swap-and-detect mechanism for post-release ownership verification that preserves task performance.
Reducing visual input to one token per frame in VLA world models maintains or improves long-horizon performance on MetaWorld, LIBERO, and real-robot tasks.
OA-WAM uses persistent address vectors and dynamic content vectors in object slots to enable addressable world-action prediction, improving robustness on manipulation benchmarks under scene changes.
CoRAL lets LLMs act as adaptive cost designers for motion planners while using VLM priors and online identification to handle unknown physics, achieving over 50% higher success rates than baselines in unseen contact-rich robotic scenarios.
VUDA enables spatial sharing between CUDA and Vulkan on GPUs via channel redirection and page-table grafting, achieving up to 85% higher throughput than temporal baselines in embodied AI tasks.
A multimodal transformer generates and caches interleaved text-image traces to guide closed-loop actions, achieving 92.4% success on LIBERO-Long and 95.5% average on LIBERO.
Introduces ISS and NMR as interventional metrics to diagnose causal misalignment in VLA policies and link it to generalization performance.
Being-H0.7 adds future-aware latent reasoning to direct VLA policies via dual-branch alignment on latent queries, matching world-model benefits at VLA efficiency.
Discrete diffusion policies act as natural asynchronous executors for robotics by treating action generation as iterative unmasking, yielding higher success rates and lower computation than flow-matching real-time chunking in dynamic tasks.
Libra-VLA introduces a coarse-to-fine dual-system architecture for VLA models that decouples discrete macro-directional planning from continuous micro-pose refinement, with performance peaking at balanced learning difficulty.
SpecRLBench is a new benchmark evaluating generalization of LTL-guided RL methods across navigation and manipulation domains with static/dynamic environments and varied robot dynamics.
CF-VLA uses a coarse initialization over endpoint velocity followed by single-step refinement to achieve strong performance with low inference steps on CALVIN, LIBERO, and real-robot tasks.
citing papers explorer
-
Beyond World-Frame Action Heads: Motion-Centric Action Frames for Vision-Language-Action Models
MCF-Proto adds a motion-centric local action frame and prototype parameterization to VLA models, inducing emergent geometric structure and improved robustness from standard demonstrations alone.
-
Thinking in Text and Images: Interleaved Vision--Language Reasoning Traces for Long-Horizon Robot Manipulation
A multimodal transformer generates and caches interleaved text-image traces to guide closed-loop actions, achieving 92.4% success on LIBERO-Long and 95.5% average on LIBERO.
-
Mini-BEHAVIOR-Gran: Revealing U-Shaped Effects of Instruction Granularity on Language-Guided Embodied Agents
Mini-BEHAVIOR-Gran benchmark reveals a U-shaped effect of instruction granularity on embodied agent performance, with planning-width correlating best and coarse instructions linked to vision-dominant shallow policies.
-
RL-VLA$^3$: A Flexible and Asynchronous Reinforcement Learning Framework for VLA Training
RL-VLA³ is an asynchronous RL framework for VLA training that delivers up to 85.2% higher throughput than synchronous baselines while preserving identical sample efficiency and scaling to 256 GPUs.
-
VLA-Trace: Diagnosing Vision-Language-Action Models through Representation and Behavior Tracing
VLA-Trace diagnoses two VLA models via representation CKA, attention interventions, and behavioral tests, finding distinct finetuning dynamics, different routing, and strong visual grounding but weak fine-grained semantic following.
-
Reinforcing VLAs in Task-Agnostic World Models
RAW-Dream disentangles world-model learning from task data by using a pre-trained task-agnostic world model and VLM rewards, with dual-noise filtering, to enable zero-shot VLA adaptation in simulation and real settings.
-
When to Re-Commit: Temporal Abstraction Discovery for Long-Horizon Vision-Language Reasoning
Learns state-conditioned commitment depth in a 7B vision-language policy that jointly predicts actions and replan intervals, outperforming fixed-depth baselines and larger models on Sliding Puzzle and Sokoban while providing a theoretical dominance result.
-
PRTS: A Primitive Reasoning and Tasking System via Contrastive Representations
PRTS pretrains VLA models with contrastive goal-conditioned RL to embed goal-reachability probabilities from offline data, yielding SOTA results on robotic benchmarks especially for long-horizon and novel instructions.
-
Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning
Single-stage fine-tuning of a video model to generate actions as latent frames plus future states and values yields state-of-the-art robot policy performance on LIBERO, RoboCasa, and bimanual tasks.
-
vla-eval: A Unified Evaluation Harness for Vision-Language-Action Models
vla-eval decouples VLA model inference from benchmark execution via WebSocket and Docker, supporting 14 benchmarks with up to 47x speedup and reproducing published scores across six codebases.
-
VGAS: Value-Guided Action-Chunk Selection for Few-Shot Vision-Language-Action Adaptation
VGAS uses best-of-N selection with a geometrically grounded critic and explicit regularization to improve success rates of few-shot VLA policies under limited data and distribution shifts.