Lukhnos Liu c8f65580bb Make UserPhrasesLM more tolerant ... This lets UserPhrasesLM consumes as much user data as possible before bailing. This makes it more tolerant to data errors and will not fail entirely just because the user has one faulty line in a data file. Also removes FastFM from the benchmarking suite. This also runs the CMake-based C++ tests as part of the GitHub CI.		2022-01-18 16:20:25 -08:00
..
ISSUE_TEMPLATE	Updates the issue templates.	2022-01-15 22:25:33 +08:00
workflows	Make UserPhrasesLM more tolerant	2022-01-18 16:20:25 -08:00