vocab=41 | params=175,337 (0.18M) | device=cpu
step     1 | loss 3.7766 | 0s
step   300 | loss 0.6712 | 13s
step   600 | loss 0.4244 | 26s
step   900 | loss 0.3414 | 40s
step  1200 | loss 0.3563 | 54s
step  1500 | loss 0.2990 | 67s
step  1800 | loss 0.1755 | 80s
step  2100 | loss 0.1144 | 93s
step  2400 | loss 0.0890 | 107s
step  2700 | loss 0.0759 | 121s
step  3000 | loss 0.0526 | 135s
step  3300 | loss 0.0262 | 148s
step  3600 | loss 0.0212 | 162s
step  3900 | loss 0.0171 | 176s
step  4200 | loss 0.0197 | 191s
step  4500 | loss 0.0173 | 206s
step  4800 | loss 0.0125 | 221s
step  5100 | loss 0.0160 | 236s
step  5400 | loss 0.0247 | 250s
step  5700 | loss 0.0046 | 265s
step  6000 | loss 0.0038 | 280s
step  6300 | loss 0.0015 | 295s
step  6600 | loss 0.0076 | 310s
step  6900 | loss 0.0006 | 324s
step  7200 | loss 0.0140 | 339s
step  7500 | loss 0.0114 | 353s
step  7800 | loss 0.0037 | 368s

[eval] exact match: 500/500 (100.0%)

[samples]
  O  '1 Oct 2032'             -> '2032-10-01'   (answer '2032-10-01')
  O  '07/06/1963'             -> '1963-07-06'   (answer '1963-07-06')
  O  '12/12/2002'             -> '2002-12-12'   (answer '2002-12-12')
  O  '05/14/2036'             -> '2036-05-14'   (answer '2036-05-14')
  O  '6 Apr 1951'             -> '1951-04-06'   (answer '1951-04-06')
  O  '2009년 3월 7일'            -> '2009-03-07'   (answer '2009-03-07')
  O  '08/27/1960'             -> '1960-08-27'   (answer '1960-08-27')
  O  '13 Jul 2000'            -> '2000-07-13'   (answer '2000-07-13')

total time: 386s
