This leverages the new logging format in that we don't need to buffer the training data, we can just write it out. Differential Revision: https://reviews.llvm.org/D142168
2.2 KiB
2.2 KiB