May 19 02:25 Update README.md 02:24 Update README.md May 16 22:15 Update README.md Apr 29 03:35 Update README.md Apr 27 18:07 Upload 2 files Apr 26 23:07 Update README.md 23:03 Rename kingpt_sf-selfplay-only_500k-iters.pt to kingpt_beaver_500k-iters_massivelyoverfit.pt 23:03 Rename kingpt_puzzles-only_500k-iters.pt to kingpt_woodpecker_500k-iters.pt 23:03 Rename kingpt_puzzles-only_1m-iters.pt to kingpt_woodpecker_1m-iters.pt 23:02 Delete kingpt-selfplay_meta.pkl 23:02 Rename kingpt-puzzles_meta.pkl to meta.pkl Apr 21 04:21 Rename kingpt-selfplay-500k_meta.pkl to kingpt-selfplay_meta.pkl I forgor (joewarida) that training a model for an additional 500k iterations does not change its vocab file
so I can actually give them good names that are concise and relevant
hurray! 04:20 Rename kingpt-puzzles-1m_meta.pkl to kingpt-puzzles_meta.pkl 04:20 Delete kingpt-puzzles-500k_meta.pkl Apr 11 01:59 Add 1m iter versions of KINGPT trained on puzzles only Apr 7 17:42 Add 500k checkpoints for KINGPT corpus variants 500k iters, corpus with puzzles only
500k iters, corpus with SF selfplay games only 17:41 initial commit 
