A graph-grammar cheminformatics workflow expands the carbon fixation reaction network, then uses integer linear programming flow queries to surface short autocatalytic pathways producing Acetyl-CoA and Malate with efficiencies approaching the CETCH cycle.
Property-optimizing ACSESS combines diversity-biased sampling with iterative fitness thresholding to discover diverse sets of molecules with favorable properties. Tested on GDB-9 (dipole moment optimization) and NKp fitness landscapes, it outperforms standard genetic algorithms in diversity while matching or exceeding their fitness, using only ~30,000 evaluations to navigate a 300,000-molecule space.
Ye et al. find that language model loss on each domain follows an exponential function of training mixture proportions. By nesting data mixing laws with scaling laws for steps and model size, small-scale experiments can predict and optimize mixtures for large models, achieving 48% training efficiency gains.
Xie et al. propose DoReMi, which trains a 280M proxy model using Group DRO to find optimal domain mixture weights, then uses those weights to train an 8B model 2.6x faster with 6.5% better downstream accuracy.
Muennighoff et al. train 400+ models to study how data repetition affects scaling. They propose a data-constrained scaling law with exponential decay for repeated tokens, finding that up to 4 epochs have negligible impact on loss, returns diminish around 16 epochs, and code augmentation provides a 2x effective data boost.
Shen et al. empirically analyze how different domain combinations and deduplication strategies in the SlimPajama dataset affect 1.3B model performance. Global deduplication across sources outperforms local deduplication, and increasing domain diversity consistently improves average accuracy, with findings transferring to 7B scale.
Raffel et al. introduce T5, a unified text-to-text framework for NLP transfer learning. Through systematic ablation of architectures, pre-training objectives, datasets, and multi-task mixing strategies, they identify best practices and scale to 11B parameters, achieving state-of-the-art results across multiple benchmarks.
Introduces Coscientist, a GPT-4-driven AI system that autonomously designs and executes chemical experiments using web search, code execution, and robotic lab automation.
The Grammar VAE replaces character-level decoding with context-free grammar production rules, using a stack-based masking mechanism to guarantee that all generated SMILES strings are syntactically valid. Applied to molecular optimization and symbolic regression, it learns smoother latent spaces and finds better molecules than character-level baselines.
Proposes Augmented Hill-Climb, a hybrid RL strategy for SMILES-based generative models that improves sample efficiency ~45-fold over REINVENT by filtering low-scoring molecules from the loss computation, with diversity filters to prevent mode collapse.
Shows that divergence between optimization and control scores during goal-directed molecular generation is explained by pre-existing disagreement among QSAR models on the training distribution, not by algorithmic exploitation of model-specific biases.
An evolutionary molecular design framework that evolves ECFP fingerprint vectors using a genetic algorithm, reconstructs valid SMILES via an RNN decoder, and evaluates fitness with a DNN property predictor.