Compositional Sculpting of Iterative Generative Processes

Authors: Timur Garipov, Sebastiaan De Peuter, Ge Yang, Vikas Garg, Samuel Kaski, Tommi Jaakkola

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We offer empirical results on image and molecular generation tasks. and 6 Experiments.
Researcher Affiliation Collaboration Timur Garipov1 Sebastiaan De Peuter2 Ge Yang1,4 Vikas Garg2,5 Samuel Kaski2,3 Tommi Jaakkola1 1MIT CSAIL 2Aalto University 3University of Manchester 4Institute for Artificial Intelligence and Fundamental Interactions 5Yai Yai Ltd
Pseudocode Yes Algorithm A.1 Compositional Sculpting: classifier training
Open Source Code Yes Project codebase: https://github.com/timgaripov/compositional-sculpting.
Open Datasets Yes three diffusion models trained to generate MNIST [56] digits {0, 1, 2, 3} in two colors: cyan and beige. and [56] Yann Le Cun. The mnist database of handwritten digits. http://yann.lecun.com/exdb/mnist/, 1998.
Dataset Splits No The paper mentions training steps and batch sizes but does not specify explicit train/validation/test dataset splits with percentages or counts.
Hardware Specification Yes All models were trained with a single Ge Force RTX 2080 Ti GPU. and All models were trained with a single Tesla V100 GPU.
Software Dependencies No The paper mentions using PyTorch [68] and optimizers like Adam [59] and Ada Delta [67] but does not provide specific version numbers for these software components (e.g., 'PyTorch 1.x' or 'Python 3.x').
Experiment Setup Yes We used Adam optimizer [59] with learning rate 0.001, and pre-train the base models for 20 000 steps with batch size 16 (16 trajectories per batch). and The score model was trained using Adam optimizer [59] with a learning rate decreasing exponentially from 10 2 to 10 4. We performed 200 training steps with batch size 32.