NanoFlow: Scalable Normalizing Flows with Sublinear Parameter Complexity
Authors: Sang-gil Lee, Sungwon Kim, Sungroh Yoon
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments performed on audio and image models confirm that our method provides a new parameter-efficient solution for scalable NFs with significant sublinear parameter complexity. |
| Researcher Affiliation | Academia | Sang-gil Lee Sungwon Kim Sungroh Yoon Data Science & AI Lab. Seoul National University {tkdrlf9202, ksw0306, sryoon}@snu.ac.kr |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide an explicit statement about releasing source code or a link to a code repository for the methodology described. |
| Open Datasets | Yes | For the performance evaluation of waveform generation, we used the LJ speech dataset [14], which is a 24-h single-speaker speech dataset containing 13,100 audio clips. ... We trained Glow, Nano Flow, and its ablations on the CIFAR10 dataset for 3,000 epochs, where all model configurations reached saturation in performance. |
| Dataset Splits | No | The paper states, 'We used the first 10% of the audio clips as the test set and the remaining 90% as the training set.' It does not explicitly mention a separate validation set split, although it refers to 'checkpoint averaging' as part of the training process. |
| Hardware Specification | Yes | using a single Nvidia V100 GPU with half-precision arithmetic. |
| Software Dependencies | No | The paper mentions using the 'Adam optimizer [16]' and refers to an 'official Wave Glow implementation [26]', but it does not specify version numbers for these or other software components. |
| Experiment Setup | Yes | We trained all models for 1.2 M iterations with a batch size of eight and an audio clip size of 16,000, using an Nvidia V100 GPU. We used the Adam optimizer [16] with an initial learning rate of 10 3, and we annealed the learning rate by half for every 200 K iterations. |