Research map / Multi-agent systems and games / LLM agents
De novo Drug Design using Reinforcement Learning with Multiple GPT Agents
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use multiple GPT agents collaborating in chemical-space search | MolRL-MGPT · GuacaMol, 20 tasks | total score: 17.983 -> 18.052 | Helped |
| Add the encouragement term for agents to explore different directions (ablation) | MolRL-MGPT · GSK3β and JNK3 top-100 | IntDiv: 0.285 ± 0.023 -> 0.362 ± 0.015; 0.345 ± 0.025 -> 0.372 ± 0.025 | Helped |
| Add experience replay (ablation) | MolRL-MGPT · GSK3β and JNK3 top-100 | mean score: 0.964 ± 0.005 -> 1.000 ± 0.000; 0.918 ± 0.008 -> 0.961 ± 0.010 | Helped |
| Use a decreasing schedule for σ1 (ablation) | MolRL-MGPT · GSK3β and JNK3 top-100 | mean score: 0.997 ± 0.001 -> 1.000 ± 0.000; 0.940 ± 0.014 -> 0.961 ± 0.010 | Helped |
| Add similarity penalization (ablation) | MolRL-MGPT · GSK3β and JNK3 top-100 | mean score: 1.000 ± 0.000 -> 1.000 ± 0.000; 0.961 ± 0.010 -> 0.956 ± 0.009 | No clear effect |
| Use 1, 2, or 8 agents instead of 4 (ablation) | MolRL-MGPT · GSK3β and JNK3 top-100 | IntDiv: GSK3β: 0.318 ± 0.020, 0.335 ± 0.017, 0.360 ± 0.020 vs 0.362 ± 0.015; JNK3: 0.343 ± 0.017, 0.357 ± 0.028, 0.369 ± 0.018 v | Mixed |
Closest papers by meaning
- Searching for High-Value Molecules Using Reinforcement Learning and Transformers (ICLR 2024 · similarity 0.85)
- MAC-AMP: A Closed-Loop Multi-Agent Collaboration System for Multi-Objective Antimicrobial Peptide Design (ICLR 2026 · similarity 0.83)
- Genetic-guided GFlowNets for Sample Efficient Molecular Optimization (NeurIPS 2024 · similarity 0.82)
- GenMol: A Drug Discovery Generalist with Discrete Diffusion (ICML 2025 · similarity 0.82)
- MolCRAFT: Structure-Based Drug Design in Continuous Parameter Space (ICML 2024 · similarity 0.82)
- Uncertainty-Aware Multi-Objective Reinforcement Learning-Guided Diffusion Models for 3D De Novo Molecular Design (NeurIPS 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.