-
Together AI Releases DeepSWE: A Fully Open-Source RL-Trained Coding Agent Based on Qwen3-32B and Achieves 59% on SWEBench
Together AI Releases DeepSWE: A Fully Open-Source RL-Trained Coding Agent Based on Qwen3-32B Achieving 59% on SWEBench Together AI has launched DeepSWE, a fully open-sourced software engineering agent trained using reinforcement learning (RL). This agent is built on the Qwen3-32B language model and has achieved 59% accuracy on the SWEBench-Verified benchmark, with a 42.2% Pass@1… →
-
Geneva Homologous Recombination Deficiency Test Is Predictive of Survival Benefit From Olaparib and Bevacizumab Maintenance in Ovarian Cancer
CONCLUSION: The Geneva HRD test predicts PFS and OS benefit from olaparib + bevacizumab. The potential detrimental effect of olaparib + bevacizumab on OS in the HRD-negative population is hypothesis-generating and needs to be confirmed prospectively. →
-
VIVALDI-CT shaping care home COVID-19 testing policy: A pragmatic cluster randomised controlled trial of asymptomatic testing compared to standard care in care home staff
CONCLUSION: This trial was not well-powered to evaluate the impact of the intervention on the primary outcome, and recorded uptake of staff testing was low. However, our pre-existing care home network underpinned by linked routinely collected data provides a model for more agile interventional studies in the care home setting. →
-
Shanghai Jiao Tong Researchers Propose OctoThinker for Reinforcement Learning-Scalable LLM Development
«`html Shanghai Jiao Tong Researchers Propose OctoThinker for Reinforcement Learning-Scalable LLM Development Introduction: Reinforcement Learning Progress through Chain-of-Thought Prompting Large Language Models (LLMs) have demonstrated significant advancements in complex reasoning tasks through Chain-of-Thought (CoT) prompting combined with large-scale reinforcement learning (RL). Models such as Deepseek-R1-Zero have exhibited strong reasoning capabilities by applying RL directly to… →
-
ReasonFlux-PRM: A Trajectory-Aware Reward Model Enhancing Chain-of-Thought Reasoning in LLMs
«`html Understanding the Role of Chain-of-Thought in LLMs Large language models (LLMs) are increasingly utilized to tackle complex tasks such as mathematics and scientific reasoning through structured chain-of-thought approaches. These models do not simply provide answers; they reason through intermediate steps that simulate logical thought processes. This technique enhances reasoning accuracy and facilitates clearer error… →
-
Effect of the Otago Exercise Program on Activities of Daily Living in Older Adults With Sarcopenia
CONCLUSION: OEP significantly enhances activities of daily living, improves muscle strength, balance, and physical function in older adults, and is more effective than conventional rehabilitation exercise programs, making it suitable for extensive clinical application. →
-
Psychoeducation for Caregivers of Individuals With Borderline Personality Disorder: A Randomized Controlled Trial of Multiple Family Group Therapy
Multiple family group (MFG) therapy is a psychoeducational intervention designed for caregivers of individuals with borderline personality disorder (BPD). This study is the first to compare MFG therapy with treatment as usual (TAU). The objectives were to assess MFG therapy’s impact on caregivers’ social functioning, perceived support from mental health professionals, coping and communication skills… →
-
Effectiveness of the WeChat-based mini-program («Sober Time ACT») on individuals with hazardous drinking in China: a randomized controlled trial
CONCLUSIONS: These findings provide preliminary prospective evidence that, despite the limitation of high dropout rate, the «Sober Time ACT» mini-program not only effectively reduces alcohol craving and consumption but also improves related psychological indicators to a certain extent, providing an innovative and effective intervention tool for addressing hazardous drinking in China. →
-
The study on delirium side-effects after ultra-brief pulse electroconvulsive therapy
CONCLUSION: Compared with BP ECT, UBP ECT may reduce the incidence of delirium post-treatment in SCZ without a significant difference in efficacy. The higher cholinesterase levels in the UBP group suggest that UBP ECT may reduce neuronal asynchronous depolarization, cholinergic disorder, and pro-inflammatory responses, thereby reducing the impact on delirium. These findings provide partial scientific… →
-
Homologous recombination deficiency (HRD) tests for ovarian cancer: a multicenter French phase II study (HERO)
BACKGROUND: The identification of homologous recombination deficient (HRD) tumor is now a crucial step for the therapeutic management of ovarian cancer. The HRD tumors are both sensitive to olaparib maintenance treatment and to platinum-based chemotherapy. Despite the large amount of HRD tests currently available, only a few HRD tests were prospectively validated on a clinical… →
