Molecular property prediction is fundamental to drug discovery and materials science, yet existing models suffer from limited interpretability and poor cross-task generalization. Without explainable chemical rationales, accurate predictions alone are insufficient for guiding costly downstream experiments. We propose MppReasoner, a multimodal large language model that incorporates verifiable chemical reasoning for molecular property prediction. Built on Qwen2.5-VL-7B-Instruct, MppReasoner integrates 2D molecular images with SMILES strings through a two-stage training strategy: supervised fine-tuning (SFT) on 16,000 expert-guided reasoning trajectories, followed by Reinforcement Learning from Principle-Guided Rewards (RLPGR). RLPGR applies rule-based rewards verified by RDKit to evaluate chemical principle usage, structural analysis, and logical consistency, providing objective feedback without human preference labels. Experiments across 8 benchmark datasets show that MppReasoner outperforms the strongest baseline by 2.95% and 5.49% on in-distribution and out-of-distribution tasks, respectively. Beyond predictive accuracy, MppReasoner generates chemically grounded reasoning traces, with expert evaluation confirming high reasoning quality across three assessed dimensions. Code is available at https://github.com/Jesse-zjx/MPPReasoner.

Designing proteins that satisfy natural language functional requirements is a central goal in protein engineering. A straightforward baseline is to fine-tune generic instruction-tuned LLMs as direct text-to-sequence generators, but this is data- and compute-hungry. With limited supervision, LLMs can produce coherent plans in text yet fail to reliably realize them as sequences. This plan–execute gap motivates ProtoCycle, an agentic framework for protein design that uses LLMs primarily to drive a multi-round, feedback-driven decision cycle. ProtoCycle couples an LLM planner with a lightweight tool environment designed to emulate the iterative workflow of human protein engineers and uses LLM-driven reflection on tool feedback to revise plans. Trained with supervised trajectories and online reinforcement learning, ProtoCycle achieves strong language alignment while maintaining competitive foldability, and ablations show that reflection substantially improves sequence quality.

Recent breakthroughs in Large Language Models (LLMs) have revolutionized scientific literature analysis. However, existing benchmarks fail to adequately evaluate the proficiency of LLMs in this domain, particularly in scenarios requiring higher-level abilities beyond mere memorization and the handling of multimodal data.In response to this gap, we introduce SciAssess, a benchmark specifically designed for the comprehensive evaluation of LLMs in scientific literature analysis. It aims to thoroughly assess the efficacy of LLMs by evaluating their capabilities in Memorization (L1), Comprehension (L2), and Analysis & Reasoning (L3). It encompasses a variety of tasks drawn from diverse scientific fields, including biology, chemistry, material, and medicine.To ensure the reliability of SciAssess, rigorous quality control measures have been implemented, ensuring accuracy, anonymization, and compliance with copyright standards. SciAssess evaluates 11 LLMs, highlighting their strengths and areas for improvement. We hope this evaluation supports the ongoing development of LLM applications in scientific literature analysis.SciAssess and its resources are available at https://github.com/sci-assess/SciAssess.

Dense retrieval requires high-quality text sequence embeddings to support effective search in the representation space. Autoencoder-based language models are appealing in dense retrieval as they train the encoder to output high-quality embedding that can reconstruct the input texts. However, in this paper, we provide theoretical analyses and show empirically that an autoencoder language model with a low reconstruction loss may not provide good sequence representations because the decoder may take shortcuts by exploiting language patterns. To address this, we propose a new self-learning method that pre-trains the autoencoder using a weak decoder, with restricted capacity and attention flexibility to push the encoder to provide better text representations. Our experiments on web search, news recommendation, and open domain question answering show that our pre-trained model significantly boosts the effectiveness and few-shot ability of dense retrieval models. Our code is available at https://github.com/microsoft/SEED-Encoder/.