4arXiv cs.CL (Computation and Language)·
FukuyamaBench and mechanism-aware training improve LLM chemical reaction reasoning
Researchers introduce a large-scale reaction mechanism reasoning dataset and FukuyamaBench, a benchmark derived from Fukuyama's Advanced Organic Reaction Mechanism textbook, to evaluate step-by-step chemical mechanism inference in LLMs. A fine-tuned Qwen3-30B-A3B model achieves 8.3% exact pathway match on FukuyamaBench Set A, outperforming the specialized FlowER model at 5.1%. The work argues that mechanism-aware training addresses hallucinations and physical inconsistencies in current chemical LLMs, which typically focus only on coarse-grained product prediction and retrosynthesis.
Related guides (1)
Related events (8)
Multi-agent LLM pipeline auto-generates 14,073 chemistry reaction rules from patent corpus
Researchers present a fully automated pipeline using a multi-agent LLM framework to classify chemical reactions and generate verifiable reaction rules across 665,901 US patent reactions. The system expands a standard taxonomy from 68 to 14,073 classes without human curation, using a verification loop that tests each generated rule against the corpus. A lightweight fingerprint classifier trained on the output achieves 97.7% accuracy on unseen reactions, matching leading proprietary classifiers while extending to novel chemistries. The work demonstrates a general approach for converting generative LLMs into self-expanding symbolic rule systems.
SupraBench: First benchmark for evaluating LLMs on supramolecular chemistry reasoning
Researchers introduce SupraBench, the first benchmark designed to systematically evaluate LLMs on supramolecular chemistry tasks including binding affinity prediction, top-binder selection, solvent identification, and host-guest description. The work also releases SupraPMC, a 16M-token corpus of supramolecular chemistry articles from Europe PMC to support domain adaptation. Evaluation of broad open and proprietary LLMs reveals substantial headroom across all tasks, with domain pretraining improving in-distribution regression but creating format compliance tradeoffs. The benchmark targets a narrow but practically important scientific domain where LLM acceleration could reduce days-long dry-lab verification cycles.
Learning to Reason with LLMs
OpenAI announced a new model or capability focused on reasoning in large language models, published on September 12, 2024. The post, hosted on the OpenAI blog, describes advances in training LLMs to perform complex multi-step reasoning. This likely corresponds to the release of the o1 (formerly 'Strawberry') model series, which uses chain-of-thought reasoning trained via reinforcement learning to achieve significantly improved performance on math, science, and coding benchmarks.
SciReasoner: Multimodal foundation model for native structural reasoning across biology, chemistry, and materials science
Researchers introduce SciReasoner, a multimodal scientific foundation model that discretizes molecular coordinates, topologies, and periodic connectivities into a unified structure-aware vocabulary for reasoning over proteins, small molecules, and inorganic crystals. The model achieves state-of-the-art performance on 67 of 86 benchmarks, with notable gains in Gene Ontology prediction (F_max 0.42→0.55), retrosynthesis accuracy (0.63→0.72), and materials phase separation. Expert evaluators rated its reasoning traces as preferred or comparable to a frontier LLM in 98% of cases, suggesting meaningful interpretability alongside accuracy gains.
Study compares human and LLM active causal reasoning, finding LLMs less efficient but near human-level on conjunctive rules
A new arXiv paper investigates whether active exploration reduces the 'conjunctive handicap' in causal learning, using a blicket detector task with adult participants who could freely intervene to identify causal objects. Results show active exploration substantially improves human conjunctive causal reasoning, though conjunctive rules still require more tests than disjunctive ones. State-of-the-art LLMs approach human-level hypothesis inference accuracy but show less efficient exploration strategies and similar conjunctive-disjunctive performance gaps, raising questions about the nature of LLM causal reasoning.
ReContext: Training-free recursive evidence replay improves LLM long-context reasoning
Researchers introduce RECONTEXT, a training-free inference-time method for improving long-context reasoning in LLMs. The approach uses model-internal relevance signals to build a query-conditioned evidence pool that is replayed before final generation, without modifying the original context, external memory, or context pruning. Experiments across eight long-context datasets at 128K context length show consistent improvements on Qwen3-4B, Qwen3-8B, and Llama3-8B. The authors provide a theoretical grounding via associative memory theory, framing attention as cue-trace association and replay as trace reactivation.
CausaLab: Scalable Benchmark for Interactive Causal Discovery by LLM Agents
CausaLab is a new evaluation environment that tests LLM agents on interactive causal discovery tasks, requiring them to recover both causal graphs and structural equations from synthetic laboratory episodes governed by randomly sampled structural causal models (SCMs). The benchmark separates predictive accuracy from genuine causal understanding, revealing a persistent gap: GPT-5.2-high achieves 92% task accuracy in a 6-node observational setting but only 0.471 all-edge F1 for mechanism recovery. Mixed observation-intervention strategies improve structural fidelity, while pure intervention strategies underperform on both metrics. Premature stopping is identified as a key agent weakness, partially mitigated by prompting models to verify hypothesis-data consistency.
RA-RFT: Retrieval-Augmented Reinforcement Fine-Tuning teaches LLMs to reason by analogy
Researchers propose Retrieval-Augmented Reinforcement Fine-Tuning (RA-RFT), a post-training framework that trains a retriever to rank contexts by expected reasoning benefit rather than semantic similarity, then fine-tunes a policy model via reinforcement learning using retrieved analogous demonstrations. The key insight is that reasoning-relevant retrieval surfaces complementary solution strategies rather than superficially similar problems. On mathematical reasoning benchmarks, RA-RFT improves AIME 2025 average@32 accuracy by 7.1 and 2.8 points over GRPO for Qwen3-1.7B and Qwen3-4B respectively, suggesting reasoning-aware retrieval is orthogonal to reward design and training curriculum improvements.