LibFewShot
LibFewShot copied to clipboard
RL-VIG
LibFewShot: A Comprehensive Library for Few-shot Learning. TPAMI 2023.
Hi Thanks for sharing this code. Is it possible to use this library with images more than 3 channels for example hyper spectral images with 256 channels?
maml方法能在单个gpu上训练，但在多个gpu上平行训练会报错。具体错误如下： ```Original Traceback (most recent call last): File "/home/cougarnet.uh.edu/pyuan2/anaconda3/envs/py37/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 61, in _worker output = module(*input, **kwargs) File "/home/cougarnet.uh.edu/pyuan2/anaconda3/envs/py37/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/cougarnet.uh.edu/pyuan2/Projects/LibFewShot/core/model/backbone/conv_four.py", line...
Protonet
你好，我在Protonet网络中计算距离矩阵时乘以了0.01，发现训练准确率提升了，请问这个是正常的吗？还是说最终的准确率与距离的数量级也有关系。 "euclidean": lambda x, y: -0.01*torch.sum( torch.pow(x.unsqueeze(2) - y.unsqueeze(1), 2), dim=3,
Method Embed. 5-way 1-shot 5-way 5-shot Reported Ours Reported Ours MeTal Conv64F 52.63 52.36 70.52 70.42 ResNet12 59.64 60.54 76.20 76.88
When we run run_test.py, which model result will be used? the last or the best
- add metal (ICCV 2021) method in meta - update pre-commit-config: - https://github.com/pre-commit/pre-commit-hooks v4.0.1 -> v4.6.0 - https://github.com/psf/black 21.7b0 -> 24.4.2 - https://gitlab.com/pycqa/flake8 -> https://github.com/PyCQA/flake8 3.9.2 -> 7.1.0 - logs:...
MCL can function as an independent classifier or be plugged into global feature-based methods, such as R2D2+MCL.
add [NeurIPS 2023 | DiffKendall: A Novel Approach for Few-Shot Learning with Differentiable Kendall's Rank Correlation] method in model, including finetuning and mertic. 训练日志链接 https://github.com/HeyDemons/LibFewShot/blob/add_diffkendall/reproduce/MetaBaselineKendall/README.md
