Zhihao Jia

@XiaotaoChen The issue should have been fixed in commit bb1219898. It would be great if you can rerun your model and verify that the bug is fixed.

@knsong Thanks for your interest in TASO. We support optimizing PyTorch graphs by transforming the graph to ONNX format using `torch.onnx`. For training, please set the batch_size, which is typically...

TASO optimizes for inference performance (i.e., minimizing forward processing time). The optimized graph is mathematically equivalent to the original graph, and therefore can be used for training as well, though...

The failure should have been fixed by commit c76ee70

Thanks for the quick response. I installed triton using `pip install triton`, whose version is 2.2.