Equivariant Neural Diffusion for Molecule Generation
Authors: François Cornet, Grigory Bartosh, Mikkel Schmidt, Christian Andersson Naesseth
NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we demonstrate the benefits of END with a comprehensive set of experiments. In Section 4.1, we first display the advantages of END for unconditional generation on 2 standard benchmarks, namely QM9 (Ramakrishnan et al., 2014) and GEOM-DRUGS (Axelrod and Gomez Bombarelli, 2022). |
| Researcher Affiliation | Academia | François Cornet Technical University of Denmark frjc@dtu.dk Grigory Bartosh University of Amsterdam g.bartosh@uva.nl Mikkel N. Schmidt Technical University of Denmark mnsc@dtu.dk Christian A. Naesseth University of Amsterdam c.a.naesseth@uva.nl |
| Pseudocode | Yes | Algorithm 1 Training algorithm of END Algorithm 2 Stochastic sampling from END |
| Open Source Code | Yes | In addition to the details provided in this section, we release a public code repository with our implementation of END. |
| Open Datasets | Yes | The QM9 dataset (Ramakrishnan et al., 2014) contains 134 thousand small- and mediumsized organic molecules... GEOM-DRUGS (Axelrod and Gomez-Bombarelli, 2022) contains 430 thousand medium- and large-sized drug-like molecules... |
| Dataset Splits | No | The paper states 'We use the same data setup as in previous work (Hoogeboom et al., 2022; Xu et al., 2022)' for unconditional generation and mentions 'validation and test sets' for conditional generation, but it does not explicitly provide the specific percentages or sample counts for training, validation, and test splits within its text. |
| Hardware Specification | Yes | All experiments were run on a single GPU. The experiments on QM9 were run on a NVIDIA SM3090 with 24 GB of memory. The experiments on GEOM-DRUGS were run on NVIDIA A100 with 40 GB of memory. |
| Software Dependencies | No | The paper mentions software like 'rdkit' and 'OPENBABEL' that were used for evaluation metrics, but it does not provide specific version numbers for these or any other software dependencies. |
| Experiment Setup | Yes | For all model variants, we employ Adam with a learning rate of 10 4. We perform gradient clipping (norm) with a value of 10 on QM9, and a value of 1 on GEOM-DRUGS. QM9: 10 layers of message passing for EDM*, while the variants of END feature 5 layers of message-passing in Fφ and 5 layers in ˆxθ. For all models, we use 256 invariant and 256 equivariant hidden features, along with an RBF expansion of dimension 64 with a cutoff of 12Å for pairwise distances. We train all models for at most 1000 epochs with a batch size of 64. |