Australian National University

Conventional 3′-end single-cell and spatial transcriptomes were built to count gene expression, yet a small subset of reads carrying untemplated poly(A) tails marks where RNAs actually end. scPolyA uses those tail reads to discover polyadenylation sites de novo, then sets UMI-counting intervals from total-read coverage and reports per-cell or per-spot terminal-site usage, 3′ UTR length and intronic polyadenylation. Across eight single-cell datasets, 63.3% of its sites fell within 25 nt of PolyA_DB 3, above polyApipe, scAPAtrap, Sierra and SCAPE and close to DeepPASS-filtered SCAPTURE.

Conventional 3′-end single-cell and spatial transcriptomes were built to count gene expression, yet a small subset of reads carrying untemplated poly(A) tails marks where RNAs actually end. scPolyA uses those tail reads to discover polyadenylation sites de novo, then sets UMI-counting intervals from total-read coverage and reports per-cell or per-spot terminal-site usage, 3′ UTR length and intronic polyadenylation. Across eight single-cell datasets, 63.3% of its sites fell within 25 nt of PolyA_DB 3, above polyApipe, scAPAtrap, Sierra and SCAPE and close to DeepPASS-filtered SCAPTURE.

Whether one policy difference transfers depends on the target: Qwen3 targets gain positively while Qwen2.5 targets under a Qwen3 reference gain almost nothing, and swapping in a Qwen2.5-3B reference turns those gains positive.

Whether one policy difference transfers depends on the target: Qwen3 targets gain positively while Qwen2.5 targets under a Qwen3 reference gain almost nothing, and swapping in a Qwen2.5-3B reference turns those gains positive.

Agent skills leave persistent artifacts such as files, configurations and cached data that later sessions read and reuse, and that cross-session channel can carry malicious influence. The work characterizes three mechanisms, Instruction Propagation, Data Boundary Violation and Evidence Poisoning, and builds SkillRelayBench with 270 adversarial cases; across seven models on Codex CLI, Gemini CLI and Claude Code, attack success rates range from 46.7% to 77.4% and attack chain success rates from 35.6% to 61.1%. The runtime defense SkillRelayGuard restores artifact provenance at read time and adds pre-execution advisory review, lowering overall attack success from 69.5% to 10.9% while raising legitimate task completion from 79.6% to 87.5%.

Agent skills leave persistent artifacts such as files, configurations and cached data that later sessions read and reuse, and that cross-session channel can carry malicious influence. The work characterizes three mechanisms, Instruction Propagation, Data Boundary Violation and Evidence Poisoning, and builds SkillRelayBench with 270 adversarial cases; across seven models on Codex CLI, Gemini CLI and Claude Code, attack success rates range from 46.7% to 77.4% and attack chain success rates from 35.6% to 61.1%. The runtime defense SkillRelayGuard restores artifact provenance at read time and adds pre-execution advisory review, lowering overall attack success from 69.5% to 10.9% while raising legitimate task completion from 79.6% to 87.5%.

The authors propose G2MLP, a training-time GNN-to-MLP distillation framework that uses Ollivier–Ricci curvature to allocate supervision between prediction-level and representation-level alignment, correcting spectral underfit on sparse graphs and spectral overfit on dense graphs; the graph-free MLP student achieves the best accuracy on all six node-classification benchmarks and transfers without architectural change to Graph Transformer teachers and link prediction.

The authors propose G2MLP, a training-time GNN-to-MLP distillation framework that uses Ollivier–Ricci curvature to allocate supervision between prediction-level and representation-level alignment, correcting spectral underfit on sparse graphs and spectral overfit on dense graphs; the graph-free MLP student achieves the best accuracy on all six node-classification benchmarks and transfers without architectural change to Graph Transformer teachers and link prediction.

The authors analyze how reinforcement learning fine-tuning (RFT) reshapes exploration in vision-language-action (VLA) policies, finding that RFT contracts the overall behavior distribution while making successful trajectories more diverse, easier to elicit, and broader in task-valid solution coverage; they then propose DRIVE, which converts temporally aligned similarity among grouped rollouts under matched task conditions into a success-conditioned intrinsic reward, improving out-of-domain success on LIBERO-Plus, ManiSkill3, and RoboTwin 2.0 and raising average out-of-domain success from 64.1% to 73.3% on a dual-arm AgileX PiPER-X platform.

The authors analyze how reinforcement learning fine-tuning (RFT) reshapes exploration in vision-language-action (VLA) policies, finding that RFT contracts the overall behavior distribution while making successful trajectories more diverse, easier to elicit, and broader in task-valid solution coverage; they then propose DRIVE, which converts temporally aligned similarity among grouped rollouts under matched task conditions into a success-conditioned intrinsic reward, improving out-of-domain success on LIBERO-Plus, ManiSkill3, and RoboTwin 2.0 and raising average out-of-domain success from 64.1% to 73.3% on a dual-arm AgileX PiPER-X platform.