Bridging Geometric States via Geometric Diffusion Bridge

Authors: Shengjie Luo, Yixian Xu, Di He, Shuxin Zheng, Tie-Yan Liu, Liwei Wang

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we empirically study the effectiveness of our Geometric Diffusion Bridge on crucial real-world challenges requiring bridging geometric states. In particular, we carefully design several experiments covering different types of data, scales and scenarios, as shown in Table 2.
Researcher Affiliation Collaboration 1State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University 2Microsoft Research AI4Science 3Center for Data Science, Peking University
Pseudocode Yes Algorithm 3 Training, Algorithm 4 Training with trajectory guidance, Algorithm 5 Sampling, Algorithm 6 Sampling with trajectory guidance
Open Source Code No The code and model checkpoints will be publicly available after the submission is acceptance.
Open Datasets Yes QM9 [79] is a medium-scale dataset... Molecule3D [116] is a large-scale dataset... We adopt Open Catalyst 2022 (OC22) dataset [105]...
Dataset Splits Yes QM9 [79] ... 110k, 10k, and 11k molecules are used for train/valid/test sets respectively; (2) Molecule3D [116] ... its train/valid/test splitting ratio is 6 : 2 : 2. ... The training set consists of 45,890 catalyst-adsorbate complexes. To better evaluate the model s performance, the validation and test sets consider the in-distribution (ID) and out-of-distribution (OOD) settings which use unseen catalysts, containing approximately 2,624 and 2,780 complexes respectively.
Hardware Specification Yes All models are trained on 16 NVIDIA V100 GPU. All models are trained on 8 NVIDIA A100 GPU.
Software Dependencies No For training, we use Adam W as the optimizer... we parameterize vθ(Rt, t; R0) by extending a Graph-Transformer based equivariant network [92, 63]... We parameterize vθ(Rt, t; R0) by using Gem Net-OC [34]...
Experiment Setup Yes For training, we use Adam W as the optimizer, and set the hyper-parameter ϵ to 1e-8 and (β1, β2) to (0.9,0.999). The gradient clip norm is set to 5.0. The peak learning rate is set to 1e-4. The batch size is set to 512. The weight decay is set to 0.0. The model is trained for 500k steps with a 30k-step warm-up stage. ... The noise scale σ is set to 0.5. For inference, we use 10 time steps with the Euler solver [12].