Introduces the TVR active viewpoint-matching task and TVRBench indoor simulation benchmark, where foundation models start at low single-digit success rates but reach 51.4% after visual-action SFT and multi-turn GRPO post-training.
hub
Sat: Spa- tial aptitude training for multimodal language models
27 Pith papers cite this work. Polarity classification is still indexing.
27
Pith papers citing it
hub tools
citation-role summary
background 3
citation-polarity summary
roles
background 3polarities
background 3representative citing papers
A closed-loop self-evolving training system for spatial reasoning in MLLMs that iteratively generates QA pairs matched to the model's current capabilities via confidence feedback, achieving gains with an order of magnitude less data.
Authors create ReasonMatch-Bench and DCRL training to boost MLLM performance on wide-baseline matching, reporting gains over baselines while preserving general capabilities.
STORM teaches LVLMs to internalize spatial-temporal reasoning via bounded latent trajectories trained with generated thought videos in two stages, improving accuracy on VideoMME, MVBench and similar benchmarks while lowering inference overhead.
Fine-tuning multimodal models on a new synthetic spatial benchmark improves generative spatial compliance on real and synthetic tasks and transfers to better spatial understanding.
SCP defines a new benchmark task for predicting spatial causal outcomes beyond direct observation and shows that 23 leading models lag far behind humans on it.
RL on Qwen2-VL-2B with SAT dataset produces R1-like reasoning and 59.47% CVBench accuracy, outperforming base model by ~30% and SFT by ~2%.
Introduces PinCoT paradigm with visual reasoning anchors, builds PIN-170K dataset via automated pipeline, and trains 4B RoboPIN model via three-stage post-training to outperform 7B baselines by 12% on embodied reasoning benchmarks.
MUSE is a unified agentic harness that improves off-the-shelf MLLMs on visual spatial planning, perception, multimodal reasoning, and fine-grained discrimination benchmarks through structured execution modules and verifier-guided repair without model retraining.
VLMs fail to ground numerical values in spatial perception on new bidirectional tasks, relying on shallow cues instead of coordinate-aware representations.
GeoWorld-VLM aligns VLM image features with intermediate representations from camera-conditioned world models via fine-tuning only the encoder and projector, yielding ~4% gains on What'sUp and VSR spatial benchmarks across two VLM backbones.
MolmoAct2 is an open VLA model that outperforms baselines like Pi-05 on 7 benchmarks and whose backbone surpasses GPT-5 on 13 embodied-reasoning tasks through new datasets, specialized training, and architecture changes for lower latency.
Distilling view-consistent future views and action-outcome supervision from a generative world model into a VLM via two-stage post-training improves dynamic spatial reasoning on SAT-Real, VSI-Bench and similar benchmarks while avoiding test-time world-model cost.
SpatialStack improves 3D spatial reasoning in vision-language models by stacking and synchronizing multi-level geometric features with the language backbone.
MLLMs show a large gap in spatial mathematical reasoning compared to humans, and a new 10,000-problem dataset helps narrow it through training.
Dual Tuning is a data curation method that jointly scores training examples for benefit and for reasoning-gain to choose between reasoning and direct-answer post-training modes for multimodal LLMs.
Mull-Tokens are modality-agnostic latent tokens that enable free-form multimodal thinking and deliver up to 16% gains on spatial reasoning benchmarks.
MiMo-Embodied is a single foundation model that achieves state-of-the-art results on 17 embodied AI benchmarks and 12 autonomous driving benchmarks through multi-stage learning, curated data, and CoT/RL fine-tuning that produces positive cross-domain transfer.
Embodied-R1 uses a pointing-centric representation and reinforced fine-tuning on a 200K dataset to achieve state-of-the-art results on embodied benchmarks plus 56.2% success in SIMPLEREnv and 87.5% on real XArm tasks without task-specific training.
ViGoRL introduces visually grounded RL that anchors reasoning steps to image coordinates and uses multi-turn zooming to outperform standard RL and supervised baselines on spatial and GUI reasoning benchmarks.
Multi-SpatialMLLM integrates depth perception, visual correspondence, and dynamic perception into MLLMs via a 27M-sample MultiSPA dataset and benchmark, yielding gains on multi-frame spatial tasks.
MVoT lets multimodal models create coherent images during chain-of-thought reasoning via a token discrepancy loss, yielding competitive or better results than text-only CoT on dynamic spatial tasks.
Embodied-R1.5 is an 8B EFM achieving SOTA on 16 of 24 embodied VLM benchmarks, fine-tunable to outperform leading VLAs, with claimed zero-shot real-robot generalization.
Thinking-RFT improves Theory of Mind accuracy by 6% over SFT on shortcut-free datasets, with 10% gains on higher-order reasoning and better generalization to new domains.
citing papers explorer
-
R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model
RL on Qwen2-VL-2B with SAT dataset produces R1-like reasoning and 59.47% CVBench accuracy, outperforming base model by ~30% and SFT by ~2%.
-
Embodied-R1: Reinforced Embodied Reasoning for General Robotic Manipulation
Embodied-R1 uses a pointing-centric representation and reinforced fine-tuning on a 200K dataset to achieve state-of-the-art results on embodied benchmarks plus 56.2% success in SIMPLEREnv and 87.5% on real XArm tasks without task-specific training.