pytorch-metric-learning
pytorch-metric-learning copied to clipboard
KevinMusgrave
The easiest way to use deep metric learning in your application. Modular, flexible, and extensible. Written in PyTorch.
Calling `AccuracyCalculator(device="cpu")` instead of `AccuracyCalculator(device=torch.device("cpu"))` crashes with an unclear exception `(AttributeError: 'str' object has no attribute 'type')`. It'd be better to check if `isinstance(device, str)`, and then do `device=torch.device(device)`
If embeddings are NaN, it raises an unclear exception: ValueError: When ref_includes_query is True, the first len(query) elements of reference must be equal to query. Likewise, the first len(query_labels) elements...
Hi, I would like to use an ONNX optimized version of a model trained using your library. However, for the use case of one-shot classification I need to determine a...
If faiss package is not installed, AccuracyCalculator() initialization raises `NameError: name 'faiss' is not defined`
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [128,...
Hi, I was looking at the implementation of MPerClassSampler, and I noticed the following issue: in consecutive batches, there are often overlaps of classes used. For example, the first batch...
Hi, is it expected that the MultiSimilarityMiner will produce positive pairs that don't actually have the same label? For example one of my batches has items with the following labels...
Metadata
Owner
Metadata
The easiest way to use deep metric learning in your application. Modular, flexible, and extensible. Written in PyTorch.