fig3

From: From descriptors to machine learning interatomic potentials: a review of AI-accelerated electrocatalyst design

Figure 3. (A) PFD workflow: fine-tune a universal pre-trained model on a small, ab initio dataset to achieve material-specific, first-principles accuracy; then distill a faster, simplified model using data generated and labeled by the fine-tuned model. Reproduced with permission from Wang et al.^[29] (CC BY 4.0); no changes made; (B) Blocks iteratively sample configurations when uncertainty > σ_thr until Δ_thr is met. Adapted with permission from Schaaf et al.^[27] (CC BY 4.0); changes made: cropped to only include panel b. PFD: Pre-train (P), Fine-tune (F), Distill (D); AL: active learning; MLFF: machine learning force field; DFT: density functional theory; CC BY 4.0: Creative Commons Attribution 4.0 International.