RSGPT, a generative model pre-trained on ten billion data points, achieving state-of-the-art performance for synthesis planning, and introduces reinforcement learning to capture the relationships among products, reactants, and templates more accurately.
Abstract
Retrosynthesis planning is a crucial task in organic synthesis, and deep-learning methods have enhanced and accelerated this process. With the advancement of the emergence of large language models, the demand for data is rapidly increasing. However, available retrosynthesis data are limited to only millions. Therefore, we pioneer the utilization of the template-based algorithm to generate chemical reaction data, resulting in the production of over 10 billion reaction datapoints. A generative pretrained transformer model is subsequently developed for template-free retrosynthesis planning by pre-training on 10 billion generated data. Inspired by the strategies of large language models, we introduce reinforcement learning to capture the relationships among products, reactants, and templates more accurately. Experiments demonstrate that our model achieves state-of-the-art performance on the benchmark, with a Top-1 accuracy of 63.4%, substantially outperforming previous models. Computer-aided synthesis-planning methods have significantly assisted synthesis planning. In this work, the authors present RSGPT, a generative model pre-trained on ten billion data points, achieving state-of-the-art performance for synthesis planning
The learning-rate schedule is a consequential choice in training deep networks, yet the policies in common use are heuristic, and published comparisons are hard to read, because architecture, dataset, and budget tend to vary alongside the schedule. We study BrachistoneLR, a schedule built by mapping the vertical coordinate of the brachistochrone, the curve of fastest descent under gravity, onto the range between a peak and a floor rate. Expanding the definition shows it to be cosine annealing with the half-period set to E - 1 instead of E, the configuration a standard implementation gives when its period argument is one less than the number of epochs. The rate therefore reaches its floor at the last epoch trained rather than one epoch later, and we show this difference decays as E^-2, making it a short-horizon effect. We then benchmark six schedules over 72 runs on three image classification datasets (MNIST, Fashion-MNIST, CIFAR-10) and four architecture families (fully connected, convolutional, recurrent, residual), fixing the optimizer, data pipeline, and evaluation protocol so that only the schedule varies. Schedules that fall smoothly from peak to floor beat the constant rate and calendar-based decay by margins that grow with task difficulty, reaching 2.5 points of dataset mean on CIFAR-10. Within that leading group, BrachistoneLR, cosine annealing, and warmup-cosine lie within 0.06 accuracy points and 0.17 of a mean rank, which one seed per configuration cannot separate. BrachistoneLR is best on both residual networks and has the highest CIFAR-10 mean, and it sets no milestones, decay factor, warmup length, or restart period. We conclude that the shape of a schedule matters more than its parameterization, that the choice of whether to use a smooth schedule matters more than the choice among them, and that the terminal-rate distinction is worth attention only over short horizons.
CoPES is introduced, a cooperative coevolutionary method that decomposes the full parameter space into lower-dimensional subspaces and searches over them cooperatively to improve optimization efficiency and demonstrate an improved trade-off between memory requirements and training time for agentic LLM post-training under resource constraints.
Zhiyuan Wang, Shengcai Liu, Jiahao Wu et al.· 1 citation
This work systematize the RL-for-LLM paradigm and provides a compute-centric analysis of prominent post-training algorithmic frameworks: Proximal Policy Optimization (PPO), Group Relative Policy Optimization (GRPO), as well as their variants, and develops a taxonomy of intra- and inter-model parallelism strategies for RL-for-LLMs.
Maciej Besta, L. Schmidt, Lara Nonino et al.· 0 citations
Reinforcement learning with verifiable rewards (RLVR) is rapidly advancing the reasoning capabilities of language models, yet the optimization layer that converts reward feedback into weight-space updates remains poorly understood. Building on our prior analysis (Zhu et al., 2025), we study this missing layer through the singular structure of model weights and identify spectral inheritance: RLVR can reuse the base model's weight spectra while acquiring new behavior through changes in the associated input and output singular frames. We operationalize spectral inheritance as Isospectral Optimization (ISO), an RLVR-native, fixed-spectrum optimization framework with complementary offline and online instantiations. Offline, ISO-Merger combines the frame changes of shared-base specialists into a single fixed-spectrum model, requiring no post-merge data, rollouts, gradient updates, or on-policy distillation (OPD). It recovers complementary specialist capabilities and achieves the strongest aggregate performance among the compared data-free merging methods. Online, ISO-Optimizer applies a chosen base optimizer, including AdamW and Muon, to the frame variables while keeping the base spectra fixed. Across reasoning and coding tasks ranging from 1.5B to 8B parameters, ISO-Optimizer improves accuracy in the reported runs and reaches matched scores with substantially fewer training steps. On Qwen3-8B-Base, AdamW reaches an aggregate accuracy of 0.495 after 270 training steps. ISO-AdamW reaches the same accuracy after only 100 training steps and improves further to 0.509 after 210 training steps. Together, ISO offers a concrete answer to RLVR's missing optimization layer: rather than inheriting pre-training optimization wholesale, design post-training around the structure of reward-driven adaptation: inherit the spectrum, optimize the frames.
Large pretrained vision-language-action models achieve strong robot-manipulation performance, while compact alternatives have largely pursued efficiency by compressing the prevailing observation-to-action paradigm. We investigate whether predictive sensorimotor modeling can make more effective use of a limited parameter budget than direct observation-to-action mapping. We present PredVLA, a language-conditioned predictive-coding policy with only 0.68 million trainable network parameters and no robot-data pretraining. Its hierarchical recurrent dynamics predict visual features and proprioception, while observations influence latent state only through prediction-error-driven online inference. On LIBERO, PredVLA achieves an 86.9% mean success rate across the three short-horizon suites and 75.4% across all four suites. Under a controlled comparison using the same frozen front end, demonstrations, action decoder, and evaluation protocol, PredVLA achieves 3.7x and 7.4x the three-suite mean success rates of parameter-matched Transformer and LSTM behavior-cloning policies, respectively. A mechanism-by-mechanism transition to the recurrent behavior-cloning baseline shows that replacing the predictive pathway with direct observation input produces the largest single performance drop, accounting for approximately $70\%$ of the endpoint gap. Further ablations identify distinct contributions from training-time latent inference, test-time error regression, hierarchical timescales, and sensory prediction-error channels. Together, these results support predictive sensorimotor modeling as a strong inductive bias for compact language-conditioned robot control.
Evaluations across scientific, multimodal, agentic, and general-purpose benchmarks show that Intern-S2-Preview-397B achieves competitive or leading results in multiple settings.
Lei Bai, Jiaqi Cao, Chiyu Chen et al.· 2 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduAug 18, 2026
A new method for surgically removing training examples from a model reveals that as datasets grow, the link between what a model learns and what it produces dissolves.
What does it take to trust AI-driven HVAC optimization? Our AI Model Factory combines agents, machine learning, reinforcement learning and deterministic checks in a governed workflow designed for messy, real-world building data. The post We built an AI factory for HVAC control appeared first on GPT-Lab.
Microsoft Research Blog· microsoft.comJul 30, 2026
LLMs do not get smarter just by remembering more. EvoLib turns experience into evolving knowledge, taking reusable skills and insights that help models learn and adapt across tasks long after deployment. The post EvoLib: Turning experience into evolving knowledge appeared first on Microsoft Research.
