Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Multisample Flow Matching: Straightening Flows with Minibatch Couplings
Authors: Aram-Alexandre Pooladian, Heli Ben-Hamu, Carles Domingo-Enrich, Brandon Amos, Yaron Lipman, Ricky T. Q. Chen
ICML 2023 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We empirically investigate Multisample Flow Matching on a suite of experiments. First, we show how different couplings affect the model on a 2D distribution. We then turn to benchmark, high-dimensional datasets, namely Image Net (Deng et al., 2009). |
| Researcher Affiliation | Collaboration | 1Meta AI (FAIR) 2Center for Data Science, NYU 3Weizmann Institute of Science 4Courant Institute of Mathematical Sciences, NYU. |
| Pseudocode | Yes | Algorithm 1 Stable Coupling (Gale Shapely) and Algorithm 2 Heuristic Coupling |
| Open Source Code | No | The paper does not provide an explicit statement or link indicating that the source code for the proposed Multisample Flow Matching method is publicly available. |
| Open Datasets | Yes | We use the official face-blurred Image Net data and then downsample to 32x32 and 64x64 using the open source preprocessing scripts from Chrabaszcz et al. (2017). Image Net (Deng et al., 2009). |
| Dataset Splits | No | The paper uses ImageNet data for experiments but does not explicitly provide details on how the dataset was split into training, validation, and test sets, or refer to specific predefined splits with citations. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU models, CPU types, or other hardware specifications used for running the experiments. |
| Software Dependencies | No | The paper mentions using 'torchdiffeq' (Chen, 2018) and the 'Tensor Flow-GAN' library, but it does not provide specific version numbers for these or other software dependencies. |
| Experiment Setup | Yes | Table 10. Hyper-parameters used for training each model. It includes details such as Channels, Depth, Batch size / GPU, GPUs, Epochs, Learning Rate, and Learning Rate Scheduler for Image Net-32 and Image Net-64. |