{"arxiv_id":"2608.13244","incoming":{"papers":0,"sentences":0,"groups":[],"unlabelled_sentences":0},"outgoing":{"papers":11,"sentences":15,"groups":[{"kind":"method_or_tool","label":"Uses the method or tool of","papers":2,"sentences":2,"quotes":[{"arxiv_id":"2308.12966","title":"Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond","url":"/paper/2308.12966","section":"Generic Patch Encoding","sentence":"The Qwen-VL vision encoder [ref] maps each depiction to patch tokens: [equation] where $Z_A,Z_B∈ℝ^N×d$; single-molecule tasks use only $Z_A$.","intent_p":0.99,"contests_p":null},{"arxiv_id":"2402.03300","title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","url":"/paper/2402.03300","section":"Training Strategy","sentence":"We apply Group Relative Policy Optimization (GRPO) [ref] to a deterministically sampled subset of SFT training records, stratified by answer outcome, property class, and action type.","intent_p":1.0,"contests_p":null}]},{"kind":"comparison","label":"Compares with","papers":1,"sentences":1,"quotes":[{"arxiv_id":"2602.07075","title":"LatentChem: From Textual CoT to Latent Thinking in Chemical Reasoning","url":"/paper/2602.07075","section":"Baselines","sentence":"The symbolic group includes instruction-based chemical LLMs, LatentChem [ref] , and Qwen3.5-4B models [ref] fine-tuned on the same reasoning data.","intent_p":0.99,"contests_p":null}]},{"kind":"background","label":"Cites as background","papers":7,"sentences":8,"quotes":[{"arxiv_id":"2602.07075","title":"LatentChem: From Textual CoT to Latent Thinking in Chemical Reasoning","url":"/paper/2602.07075","section":"Textual and Latent Chemical Reasoning","sentence":"Coconut introduces continuous reasoning for general tasks, while LatentChem applies latent computation to chemical problems [ref] .","intent_p":0.85,"contests_p":null},{"arxiv_id":"2510.10248","title":null,"url":"/paper/2510.10248","section":"Introduction","sentence":"MPPReasoner jointly processes molecular images and SMILES and rewards textual reasoning that describes tool-identified local structures [ref] .","intent_p":0.84,"contests_p":null},{"arxiv_id":"2603.12344","title":"Can Decision Trees Teach Large Language Models? Distilling Verbalized Knowledge for Molecular Property Prediction","url":"/paper/2603.12344","section":"Molecular Structure Reasoning with LLMs","sentence":"MSR provides extracted components such as functional groups as natural-language descriptions, while TreeKD converts functional-group features into textual rules for property prediction [ref] .","intent_p":0.81,"contests_p":null},{"arxiv_id":"2402.06852","title":"ChemLLM: A Chemical Large Language Model","url":"/paper/2402.06852","section":"Introduction","sentence":"Recent chemical LLMs and vision-language models (VLMs) can reason over increasingly rich molecular inputs [ref] .","intent_p":0.99,"contests_p":null},{"arxiv_id":"2409.13194","title":"ChemDFM-X: Towards Large Multimodal Model for Chemistry","url":"/paper/2409.13194","section":"Molecular Images and Local Structure","sentence":"ChemVLM and ChemMLLM adapt multimodal language models to molecular images and chemical tasks, while ChemDFM-X aligns multiple chemical modalities [ref] .","intent_p":0.99,"contests_p":null}]}],"unlabelled_sentences":4},"claims":[],"thresholds":{"intent_p":0.8,"supports_p":0.95,"contests_p":0.7},"method":"Citing sentences come from each paper's LaTeX source; every \\cite marks the sentence and the works it cites. Pith's classifier typed each sentence. A label is shown at intent probability 0.8 or more (0.95 for support), a contest at 0.7 or more. Measured accuracy against two blind model families is at https://pith.science/accuracy. The quoted sentences are the citing authors' own.","held_up":null,"_links":{"html":"https://pith.science/paper/2608.13244"}}