Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Retrosynthesis planning seeks synthesis routes from purchasable starting materials to a target molecule. Search over predicted reactions can connect a target to available materials, yet an infeasible reaction can invalidate the resulting route. Evaluating route quality only after search has finishe…
arxiv.org
October 31, 2026 at 6:52 AM
Octavian Susanu, Nguyen Xuan-Vu, Philippe Schwaller: Steerable Single-step Retrosynthesis with Electron-pushing Discrete Flow Matching and Feynman-Kac Steering https://arxiv.org/abs/2611.03696 https://arxiv.org/pdf/2611.03696 https://arxiv.org/html/2611.03696
November 4, 2026 at 6:56 AM
Octavian Susanu, Nguyen Xuan-Vu, Philippe Schwaller: Steerable Single-step Retrosynthesis with Electron-pushing Discrete Flow Matching and Feynman-Kac Steering https://arxiv.org/abs/2611.03696 https://arxiv.org/pdf/2611.03696 https://arxiv.org/html/2611.03696
Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Retrosynthesis planning seeks synthesis routes from purchasable starting materials to a target molecule. Search over predicted reactions can connect a target to available materials, yet an infeasible reaction can invalidate the resulting route. Evaluating route quality only after search has finishe…
arxiv.org
October 31, 2026 at 9:15 AM
Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Retrosynthesis planning seeks synthesis routes from purchasable starting materials to a target molecule. Search over predicted reactions can connect a target to available materials, yet an infeasible reaction can invalidate the resulting route. Evaluating route quality only after search has finishe…
arxiv.org
October 30, 2026 at 9:28 AM
Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Hongxuan Shi, Hankz Hankui Zhuo
#arXiv #cs.AI
Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Retrosynthesis planning seeks synthesis routes from purchasable starting materials to a target molecule. Search over predicted reactions can connect a target to available materials, yet an infeasible reaction can invalidate the resulting route. Evaluating route quality only after search has finishe…
arxiv.org
October 30, 2026 at 6:59 AM
Reaction precedents extend retrosynthetic prediction beyond training coverage
Jiaxing Sun et al.
#arXiv #cs.AI
Jiaxing Sun et al.
#arXiv #cs.AI
Reaction precedents extend retrosynthetic prediction beyond training coverage
Reaction precedents guide retrosynthesis, but how far they can compensate for gaps in task-specific training coverage remains unclear. Using 2.71 million reactions, we restricted training and retrieval coverage by transformation or time, then expanded only the retrieval corpus with model parameters…
arxiv.org
October 31, 2026 at 2:38 AM
Hongxuan Shi, Hankz Hankui Zhuo: Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing https://arxiv.org/abs/2610.38245 https://arxiv.org/pdf/2610.38245 https://arxiv.org/html/2610.38245
October 30, 2026 at 6:44 AM
Hongxuan Shi, Hankz Hankui Zhuo: Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing https://arxiv.org/abs/2610.38245 https://arxiv.org/pdf/2610.38245 https://arxiv.org/html/2610.38245
Wang, Wang, Wang, Feng, Zhang, Ouyang, Wu, Wang, Zheng, Huang, Zhou: Solvent: Long-Horizon Reasoning for Multi-Step Retrosynthesis https://arxiv.org/abs/2610.32308 https://arxiv.org/pdf/2610.32308 https://arxiv.org/html/2610.32308
October 26, 2026 at 6:39 AM
Wang, Wang, Wang, Feng, Zhang, Ouyang, Wu, Wang, Zheng, Huang, Zhou: Solvent: Long-Horizon Reasoning for Multi-Step Retrosynthesis https://arxiv.org/abs/2610.32308 https://arxiv.org/pdf/2610.32308 https://arxiv.org/html/2610.32308
📢 First-round molecule released! bit.ly/47dgdK6
The 14th SCI-RSC Retrosynthesis Competition is now open! This year’s target molecule is Daphcalycine B. Entries for Round one closes on 4 Dec 2026. Open to UK & Ireland.
Sponsorship opportunities available.
The 14th SCI-RSC Retrosynthesis Competition is now open! This year’s target molecule is Daphcalycine B. Entries for Round one closes on 4 Dec 2026. Open to UK & Ireland.
Sponsorship opportunities available.
October 23, 2026 at 2:59 PM
📢 First-round molecule released! bit.ly/47dgdK6
The 14th SCI-RSC Retrosynthesis Competition is now open! This year’s target molecule is Daphcalycine B. Entries for Round one closes on 4 Dec 2026. Open to UK & Ireland.
Sponsorship opportunities available.
The 14th SCI-RSC Retrosynthesis Competition is now open! This year’s target molecule is Daphcalycine B. Entries for Round one closes on 4 Dec 2026. Open to UK & Ireland.
Sponsorship opportunities available.
ReTRACE: Feedback-Guided LLM Agent in a Chemistry Harness for Retrosynthesis Planning
Jinxiang Xie et al.
#arXiv #cs.AI
Jinxiang Xie et al.
#arXiv #cs.AI
ReTRACE: Feedback-Guided LLM Agent in a Chemistry Harness for Retrosynthesis Planning
Computer-aided retrosynthesis has advanced through single-step prediction, tree search and route scoring, but complex targets still require judgement over the whole route, where stereochemistry, functional-group compatibility, cost and greenness interact. Here we present \textbf{\mbox{ReTRACE}} (\t…
arxiv.org
October 20, 2026 at 5:50 AM
ReTRACE: Feedback-Guided LLM Agent in a Chemistry Harness for Retrosynthesis Planning
Jinxiang Xie et al.
#arXiv #cs.AI
Jinxiang Xie et al.
#arXiv #cs.AI
ReTRACE: Feedback-Guided LLM Agent in a Chemistry Harness for Retrosynthesis Planning
Computer-aided retrosynthesis has advanced through single-step prediction, tree search and route scoring, but complex targets still require judgement over the whole route, where stereochemistry, functional-group compatibility, cost and greenness interact. Here we present \textbf{\mbox{ReTRACE}} (\t…
arxiv.org
October 20, 2026 at 8:10 AM
Jinxiang Xie, Yijian Zhang, Qingyuan Yi, Jianyi Wang, Hao Dong: ReTRACE: Feedback-Guided LLM Agent in a Chemistry Harness for Retrosynthesis Planning https://arxiv.org/abs/2610.24744 https://arxiv.org/pdf/2610.24744 https://arxiv.org/html/2610.24744
October 20, 2026 at 6:39 AM
Jinxiang Xie, Yijian Zhang, Qingyuan Yi, Jianyi Wang, Hao Dong: ReTRACE: Feedback-Guided LLM Agent in a Chemistry Harness for Retrosynthesis Planning https://arxiv.org/abs/2610.24744 https://arxiv.org/pdf/2610.24744 https://arxiv.org/html/2610.24744
Evaluating Machine Learning Reliability in Computer-Aided Retrosynthesis Planning: Domain Generalization and Out-of-Distribution Testing + Video
Introduction: Computer-aided retrosynthesis planning relies heavily on artificial intelligence and machine learning models to predict chemical reaction…
Introduction: Computer-aided retrosynthesis planning relies heavily on artificial intelligence and machine learning models to predict chemical reaction…
Evaluating Machine Learning Reliability in Computer-Aided Retrosynthesis Planning: Domain Generalization and Out-of-Distribution Testing + Video
Introduction: Computer-aided retrosynthesis planning relies heavily on artificial intelligence and machine learning models to predict chemical reaction pathways and molecular transformations. However, a major challenge in deploying these systems in real-world chemical research is domain generalization—specifically, how reliable predictive models remain when encountering unfamiliar molecular structures or out-of-distribution data that differ significantly from their training sets. This article explores evaluation frameworks, experimental design methodologies, and reproducible validation pipelines required to assess the robustness of AI-driven retrosynthesis tools.
undercodetesting.com
October 19, 2026 at 4:59 PM
Evaluating Machine Learning Reliability in Computer-Aided Retrosynthesis Planning: Domain Generalization and Out-of-Distribution Testing + Video
Introduction: Computer-aided retrosynthesis planning relies heavily on artificial intelligence and machine learning models to predict chemical reaction…
Introduction: Computer-aided retrosynthesis planning relies heavily on artificial intelligence and machine learning models to predict chemical reaction…
I like the word retrosynthesis. Should be used adapted to the context of thesis, antithesis, synthesis.
en.wikipedia.org/wiki/Retrosy...
en.wikipedia.org/wiki/Retrosy...
Retrosynthetic analysis - Wikipedia
en.wikipedia.org
October 17, 2026 at 4:29 PM
I like the word retrosynthesis. Should be used adapted to the context of thesis, antithesis, synthesis.
en.wikipedia.org/wiki/Retrosy...
en.wikipedia.org/wiki/Retrosy...
Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
Zixun Lan et al.
#arXiv #cs.AI
Zixun Lan et al.
#arXiv #cs.AI
Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
Single-step retrosynthesis is a one-to-many inverse problem, yet most models are trained and evaluated against a single recorded reactant set. Reference-based training gives no explicit positive signal for unrecorded alternatives, and exact-match evaluation rejects non-matching predictions regardle…
arxiv.org
October 15, 2026 at 8:41 PM
Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
Zixun Lan et al.
#arXiv #cs.AI
Zixun Lan et al.
#arXiv #cs.AI
Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
Single-step retrosynthesis is a one-to-many inverse problem, yet most models are trained and evaluated against a single recorded reactant set. Reference-based training gives no explicit positive signal for unrecorded alternatives, and exact-match evaluation rejects non-matching predictions regardle…
arxiv.org
October 15, 2026 at 10:17 PM
Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
Zixun Lan et al.
#arXiv #cs.AI
Zixun Lan et al.
#arXiv #cs.AI
Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback
Single-step retrosynthesis is a one-to-many inverse problem, yet most models are trained and evaluated against a single recorded reactant set. Reference-based training gives no explicit positive signal for unrecorded alternatives, and exact-match evaluation rejects non-matching predictions regardle…
arxiv.org
October 14, 2026 at 8:50 PM
Kim, Jeong, Yun, Ko, Oh, Lim, Kim: RetroPlanner: Multi-Route Expansion and Multi-Objective Reasoning for LLM-Based Retrosynthesis Planning https://arxiv.org/abs/2610.16391 https://arxiv.org/pdf/2610.16391 https://arxiv.org/html/2610.16391
October 15, 2026 at 6:41 AM
Kim, Jeong, Yun, Ko, Oh, Lim, Kim: RetroPlanner: Multi-Route Expansion and Multi-Objective Reasoning for LLM-Based Retrosynthesis Planning https://arxiv.org/abs/2610.16391 https://arxiv.org/pdf/2610.16391 https://arxiv.org/html/2610.16391
Zixun Lan, Zhenfu Liu, Yi Li, Xinheng Wang, Mingmin Chi, Qibiao Wang, Jie Zhu, Xiaohui Tian, Fei Ma: Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback https://arxiv.org/abs/2610.15141 https://arxiv.org/pdf/2610.15141 https://arxiv.org/html/2610.15141
October 15, 2026 at 6:40 AM
Zixun Lan, Zhenfu Liu, Yi Li, Xinheng Wang, Mingmin Chi, Qibiao Wang, Jie Zhu, Xiaohui Tian, Fei Ma: Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback https://arxiv.org/abs/2610.15141 https://arxiv.org/pdf/2610.15141 https://arxiv.org/html/2610.15141
Kim, Jeong, Yun, Ko, Oh, Lim, Kim: RetroPlanner: Multi-Route Expansion and Multi-Objective Reasoning for LLM-Based Retrosynthesis Planning https://arxiv.org/abs/2610.16391 https://arxiv.org/pdf/2610.16391 https://arxiv.org/html/2610.16391
October 14, 2026 at 6:42 AM
Kim, Jeong, Yun, Ko, Oh, Lim, Kim: RetroPlanner: Multi-Route Expansion and Multi-Objective Reasoning for LLM-Based Retrosynthesis Planning https://arxiv.org/abs/2610.16391 https://arxiv.org/pdf/2610.16391 https://arxiv.org/html/2610.16391
Zixun Lan, Zhenfu Liu, Yi Li, Xinheng Wang, Mingmin Chi, Qibiao Wang, Jie Zhu, Xiaohui Tian, Fei Ma: Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback https://arxiv.org/abs/2610.15141 https://arxiv.org/pdf/2610.15141 https://arxiv.org/html/2610.15141
October 14, 2026 at 6:40 AM
Zixun Lan, Zhenfu Liu, Yi Li, Xinheng Wang, Mingmin Chi, Qibiao Wang, Jie Zhu, Xiaohui Tian, Fei Ma: Single-step Retrosynthesis via Reinforcement Learning from Forward Feedback https://arxiv.org/abs/2610.15141 https://arxiv.org/pdf/2610.15141 https://arxiv.org/html/2610.15141
More gains, less hallucinations – how AI is taking on retrosynthesis.
Microsoft’s split-personality retrosynthetic AI is a fast learner
Retrochimera combines two deep learning algorithms from different classes, outperforming their individual success rates on datasets from Novartis and GSK
www.chemistryworld.com
October 13, 2026 at 2:08 PM
More gains, less hallucinations – how AI is taking on retrosynthesis.