Large language models (LLMs) have demonstrated strong performance in mental health analysis tasks when equipped with advanced reasoning capabilities. However, their substantial parameter sizes and high computational demands present significant barriers for routine clinical use. Recent studies have explored reasoning distillation as a means to transfer these capabilities to small language models (SLMs). However, SLMs often struggle with complex reasoning tasks due to their limited capacity to model both general cognitive abilities and specialized domain knowledge. In this paper, we propose symbolic preference distillation (SyPD), to enhance the complex reasoning abilities of SLMs in mental health analysis tasks. First, to handle challenging or ambiguous cases, we introduce a reasoning optimization strategy that leverages specialized domain knowledge to perform SLMs' error analysis and generate symbolic knowledge via a teacher. Second, to further boost SLM's reasoning ability, we propose a preference distillation method that guides an SLM to align with high-quality and clinically relevant reasoning derived from the teacher LLM through preference signals and symbolic knowledge, without requiring access to the teacher's output probabilities. By anchoring the optimization to the model's own pre aligned distribution, our method enables post-hoc correction of failure cases while gaining domain-specific knowledge. Experimental results demonstrate that our proposed SyPD, with only 1.1 billion parameters, achieves an average weighted F1-score of 0.815 on mental disorder diagnosis on the interpretable mental health instruction (IMHI) bench mark. It outperforms the state-of-the-art instruction-tuned MentaLLaMA-Chat-13B model by 6.14%, and the few-shot tuned GPT-4 model by 15.44%.
Lu Yu, Weikang Xiang, Kang Han et al.· IEEE journal of biomedical a...· 0 citations
In recent years, advances in quantum computing have been driven by substantial improvements in both the number and quality of qubits. As the field progresses, there is growing interest in interconnecting quantum systems to enable scalable computation through Distributed Quantum Computing (DQC) architectures. Consider a distributed quantum application executed across multiple Quantum Processing Units (QPUs) within a quantum data center, where remote gates require establishing entanglement between different QPUs. The creation of such end-to-end entanglement can lead to network congestion and resource contention. To address these challenges, we propose a resource management framework that maximizes fidelity-guaranteed throughput while satisfying dependency constraints. We first formulate the problem as a Mixed-Integer Linear Programming (MILP) model to provide a performance benchmark. Building on this, we develop efficient approximate scheduling algorithms that achieve performance comparable to the optimization solver. Although a trade-off exists between execution time and network throughput, simulation results demonstrate that one of the proposed strategies, Weighted Group Least Resource First (WGLRF), closely approximates the solver’s performance across most scenarios. These findings suggest that the lightweight strategy is sufficient for current DQC settings, offering a practical solution for managing remote-gate resource contention in distributed quantum circuits and improving overall system performance.
The results suggest that current VLA benchmarks may exert limited pressure on deep language grounding and compositional instruction understanding, and that future VLA architectures should allocate capacity more deliberately across language, vision, and action components.
Guoheng Sun, Kai Feng, Shwai He et al.· arXiv.org· 0 citations