Continual post-training of large multimodal models should add new capabilities while preserving those from pre-training, and the two goals pull in opposite directions. SFT gives explicit target supervision that learns a task from near-zero accuracy, but its off-policy targets move the model far enough to cause forgetting; on-policy methods such as RLVR and self-distillation preserve policy proximity yet supply little signal when the policy cannot yet solve the task. We introduce ReDraft (Reference-Driven Revision and Fine-Tuning), which obtains both from the model's own failures: using an expert response only as a reference, it has the model revise its own incorrect rollout, keeps the revision only if a verifier accepts it, and fine-tunes on what survives. Each retained target is therefore explicit, yet still close to the current policy. Across Counting, Clock Reading, and Jigsaw on Qwen2.5-VL-3B/7B, two of them with near-zero accuracy, ReDraft gains 56.9 points on the target task against SFT's 52.9 while cutting prior-task loss from 16.6 to 1.5 points (11.3x less forgetting), and improves on OPSD along both axes (19.3 gain, 6.2 loss). Data- and parameter-space analyses match the design: revised targets are more probable under the base model, and the updates they induce stay compact and follow SFT's direction more closely than OPSD's. Together, these results show that revising the model's own rollout rather than directly imitating an expert trajectory can reconcile cold-start acquisition with prior-capability retention.
Reinforcement learning with verifiable rewards (RLVR) is bottlenecked by hard prompts on which correct trajectories have low probability, so sampling misses them within a practical budget and leaves the policy update with little useful signal. We frame such zero-hit prompts as RLVR's sampling frontier, where new reasoning behavior is most valuable yet least likely to be sampled. Importantly, failed rollouts can be informative: they expose where the model's reasoning went wrong. We introduce LatentRevise, a first-order latent revision method that recovers training signal for this zero-hit regime. Given a failed rollout and the gold answer as an anchor, LatentRevise optimizes the input embeddings of its reasoning prefix under two complementary gradients, moving the prefix away from the failed continuation and toward the gold answer. The optimization is constrained to the convex hull of the model's vocabulary embeddings, so each update moves the latent toward a real token embedding rather than an arbitrary feature direction. We find that continuations from the revised prefix lengthen, exhibit self-reflection, and reach correct answers missed by the original rollouts. Used as training data, these trajectories improve SFT and RLVR on math benchmarks over standard baselines.
Test-time scaling via sequential revision has emerged as a powerful paradigm for enhancing Large Language Model (LLM) reasoning. However, standard post-training methods primarily optimize single-shot objectives, creating a fundamental misalignment with multi-step inference dynamics. While recent work treats this as multi-turn reinforcement learning (RL), conventional approaches optimize over the multi-step trajectories directly, failing to further exploit the high-quality mistakes in intermediate steps that model can learn from correcting them. We propose a two-stage iterative framework that alternates between online data/prompt augmentation and policy optimization. By converting the intermediate steps (``near-miss'' answers) in the successful recovery trajectories into decoupled revision and verification prompts, our approach concentrates training on both effective answer transformation and error identification. This approach enables efficient off-policy data generation and reduces the computational overhead of long-horizon sampling compared to standard multi-turn RL. On LiveCodeBench, using publicly available test cases as feedback, we observe gains of +6.5 points over the RL baseline and +4.0 points over standard multi-turn training. Beyond coding, our approach matches the previously reported SOTA result on circle packing while using the smallest base model (4B) and far fewer rollouts than the much larger evolutionary search systems. Math results under ground-truth verification further confirm improved correction ability. It also generalizes to out-of-distribution constraint-satisfaction puzzles such as n_queens and mini_sudoku, where correctness is defined entirely by problem constraints. Code is available at https://github.com/yxliu02/REVES.git.
Continual post-training is becoming a central paradigm for adapting vision-language models to evolving tasks. Recent work has increasingly favored reinforcement learning over supervised fine-tuning, driven by the belief that reinforcement learning is inherently less prone to forgetting. However, the belief remains insufficiently validated, as existing evidence is largely drawn from outdated or homogeneous benchmarks. We revisit this assumption under recent and diverse multimodal reasoning tasks. To this end, we introduce MRCL, a Multimodal Reasoning Continual Learning benchmark. Experiments on MRCL show that standard reinforcement learning still suffers from severe catastrophic forgetting during continual post-training. We trace this failure to an objective mismatch: the KL regularization used in common policy optimization methods is evaluated on current-task data, whereas forgetting is caused by behavioral drift on prior-task distributions. To address this problem, we propose Continual Policy Optimization (CPO), a replay-free framework grounded in a prior-task behavioral KL objective. CPO relaxes the intractable historical KL constraint into sparse parameter-movement regularization, limiting policy drift without storing old data. Extensive experiments across multiple model scales show that CPO consistently reduces forgetting while preserving, and in some cases improving, pretrained model capabilities. On Qwen3-VL-8B, CPO reduces forgetting by 13.7% and improves pretrained capability by 7.0%. The implementation code is available at https://github.com/MaolinLuo/CPO.