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].
SpikeZIP-TF: Conversion is All You Need for Transformer-based SNN
Authors: Kang You, Zekai Xu, Chen Nie, Zhijie Deng, Qinghai Guo, Xiang Wang, Zhezhi He
ICML 2024 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 4. Experiments, 4.2. Results Comparison, Table 5. Comparison on Image Net. |
| Researcher Affiliation | Collaboration | 1School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai, China 2Huawei Technologies, Shenzhen, China. |
| Pseudocode | No | The paper does not contain any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | The code is publicly available at: https://github.com/IntelligentComputing-Research-Group/SpikeZIP-TF |
| Open Datasets | Yes | Vision Benchmarks. Various vision datasets are adopted for evaluation. 1) static vision datasets, including CIFAR10/100 (Krizhevsky et al., 2009) and Image Net (Deng et al., 2009). 2) neuromorphic vision dataset: We evaluate Spike ZIP-TF on CIFAR10-DVS (Hongmin et al., 2017). |
| Dataset Splits | No | The paper lists the datasets used (e.g., Image Net, CIFAR10/100) but does not explicitly provide specific train/validation/test dataset splits (e.g., percentages or sample counts) within the main text or appendices for reproducibility. |
| Hardware Specification | No | The paper mentions 'Nvidia' GPUs and 'GPU Time(h)' in Table 7 regarding training cost, but does not specify exact GPU models (e.g., A100, V100), CPU models, or other specific hardware configurations used for running experiments. |
| Software Dependencies | No | The paper does not explicitly list specific software dependencies with version numbers (e.g., 'Python 3.8', 'PyTorch 1.9') used for the experiments. |
| Experiment Setup | Yes | The paper includes detailed hyperparameter configurations in Appendix A4, such as in Table A3 'The hyperparameter of end-to-end finetuning with Vi T-S Re LU on Image Net', listing 'optimizer Adam W', 'base learning rate 1e-4', 'batch size 64/256', 'training epochs 100', etc. |