AI Coding Live News

While supervised fine-tuning (SFT) distills agentic tool-use capabilities from frontier LLMs into small language models (SLMs, 0.6B to 8B parameters), student agents frequently suffer from 'progress stagnation'—collapsing into repetitive, unproductive tool-calling loops as interaction trajectories extend. In arXiv:2610.19439, researchers introduced Progress-Aware Distillation (ProD). ProD defines a rigorous stagnation scoring formulation that dynamically identifies and penalizes stagnant behavior during student trajectory rollouts via an adaptive SimPO-inspired margin. Experiments across 8 diverse SLMs prove ProD effectively eradicates repetitive loops while boosting task completion efficiency and success rates.

Zoom

Researchers from the University of Maryland, Google, and MBZUAI introduced BeliefWeaver (arXiv:2610.18082), an autonomous failure-recovery framework for long-horizon agents. Addressing cases where failures stem from root errors multiple steps earlier that local reflection cannot fix, BeliefWeaver maintains a persistent relational belief state across checkpoints and attempts, using an Implicit World Model (IWM) to evaluate hypotheses and gate repair execution. When distilled into a 9B model via on-policy distillation, it improves final benchmark scores by 9.8 percentage points while slashing teacher API costs by 54.8%.

Zoom

Three new frameworks advance web and multimodal agent post-training: Salesforce AI Research's CLIFT (arXiv:2610.06829) certifies URL-conditional self-verification questions for dense RL rewards and judge-free Conformal Trajectory Selection (CTS), propelling Gemma-4 31B to a 74.6% SOTA on WebArena Infinity (beating Gemini 3 Flash at 70.1%) and lifting GPT-5.5 to 53.7% on VisualWebArena. Meanwhile, MMSearch-OPD (arXiv:2610.22240) boosts 9B multimodal deep search by +15.3 points to 60.2%, and Northeastern University's SCOPED-8B (arXiv:2610.22217) uses skill-decoupled on-policy expert distillation to reach 73.13% across 8 benchmarks—surpassing Qwen3-VL-32B (72.01%) and Claude Opus 4.6 (72.86%).

Zoom

Topping today's Hugging Face Daily Papers are Waypoint-OPD (451 upvotes, arXiv:2610.21601) and FDE-Gym (230 upvotes, arXiv:2610.22350). Waypoint-OPD leverages an intermediate RL teacher checkpoint to gate endpoint policy-shift distillation on Qwen3-8B, 14B, and 30B-A3B students, achieving #1 accuracy across all six math and coding settings. Meanwhile, FDE-Gym benchmarks 13 frontier agent configurations across 95 realistic forward-deployed engineering delivery tasks, revealing that the best system resolves only 14 of 95 tasks when unstated customer constraints must be uncovered.

Zoom

To improve credit assignment in long-horizon agentic tasks, combining reinforcement learning with verifiable rewards (RLVR) and on-policy self-distillation (OPSD) has emerged as a leading paradigm. However, existing pipelines neglect the dynamic evolution of agentic skills and the reliability of privileged teacher signals. Researchers from CUHK and Tencent introduced RSIPO, enabling skills and policies to co-evolve recursively. RSIPO extracts skills directly from trajectory rollouts, optimizes skill generation via replay, and employs Empirical Bayes shrinkage to calibrate student trust in teacher logits. Operating on Qwen2.5-7B, RSIPO achieves 98.6% success on ALFWorld, outperforming standard GRPO by 17.4 percentage points.

Zoom

Salesforce AI Research has introduced Prospective Hindsight (PH), a self-calibrating training framework for long-horizon autonomous agents. Addressing the fundamental limitation of standard RL where credit assignment occurs strictly after environmental feedback, PH exploits the gap between the agent's pre-action prospective prediction and post-action retrospective evaluation. By weighting gradients with this surprise residual, PH eliminates overconfident failures in single-turn tasks and underconfident successes in multi-turn interactions across GRPO and on-policy distillation.

Zoom

On-policy distillation (OPD) is critical for reasoning and agent alignment, yet conventional methods realize only a single continuation per decision, severely restricting supervision under finite compute budgets. Researchers have introduced BranchOPD (Branching On-Policy Distillation), an algorithm that dynamically identifies high-uncertainty and teacher-student disagreement decision nodes to branch and resample continuations from shared prefixes. BranchOPD slashes generation costs by 19%–31% on agent benchmarks while boosting math reasoning accuracy by 18.55 percentage points.

Zoom

Reinforcement learning for language model agents is constrained by sparse episodic rewards and inefficient surface token alignment. Researchers have introduced PR-OPD (Privileged Representation On-policy Self-Distillation), discovering that injecting hindsight skills alters less than 25% of token probabilities while reconfiguring over 80% of internal hidden states. By aligning multi-layer projected representations against a stop-gradient self-teacher, PR-OPD boosts WebShop accuracy by 14.0 points and ALFWorld by 4.7 points with zero inference overhead.

Zoom

Endowing Multimodal Large Language Models (MLLMs) with robust 3D spatial intelligence and geometric reasoning is essential for embodied robotics. Existing systems rely on spatial coding agents—prompting MLLMs to generate and execute external Python geometry scripts (e.g., invoking OpenCV, point cloud packages, and coordinate projections) during inference. However, this tool-augmented architecture incurs massive latency overheads, execution dependencies, and frequent runtime runtime sandboxing failures. In newly released research (arXiv:2610.11366), researchers introduce SpatialOPSD (code: github.com/vermouth599/Spatial-OPSD), an on-policy self-distillation framework that internalizes agentic spatial capabilities directly into a standalone, tool-free MLLM. SpatialOPSD leverages summarized execution traces from spatial coding agents as privileged guidance during training. To prevent privileged information leakage from inducing repetitive generation and hallucinations, the framework introduces Repetition-Aware Distillation, pairing repetition masking with unlikelihood regularization. Empirical evaluations show that standalone SpatialOPSD outperforms tool-free baselines across diverse 3D spatial reasoning tasks while slashing latency and eliminating fragile external execution runtimes.

Zoom

Reinforcement Learning with Verifiable Rewards (RLVR) powers modern reasoning LLMs by translating post-interaction scalar outcomes into optimization signals. However, in group-relative objectives (e.g., GRPO), reward signals vanish completely when all rollouts within a group share the same score (such as widespread mutual failures), discarding rich diagnostic feedback regarding task requirements and failure modes. In smaller models (such as 2B parameters), up to 98% of rollout groups result in universal failure, leaving standard RLVR stranded at 0% task success. Salesforce AI Research introduces Self-Retrospection Distillation (SRD, arXiv:2610.08077, repo: github.com/SalesforceAIResearch/SRD). SRD instantiates prospective learning, using privileged post-hoc trajectories to supervise pre-interaction foresight representations within the same policy without adding inference latency. Across 10 tool-integrated reasoning and long-horizon agentic benchmarks, SRD delivers gains of up to 24.2 percentage points over baseline RLVR and self-distillation. Critically, in 2B model regimes where 98% of groups are all-failure and standard RLVR flatlines at 0.0%, adding SRD reaches a 60.6% success rate under identical rollout budgets.

Zoom