ffm模型训练的时候Precision为nan

[sytresbien]

[------------] Epoch Train log_loss Test log_loss Test Precision Time cost (sec) [ 10% ] 1 0.087384 0.087865 -nan(ind) 1.46 [ 20% ] 2 0.087555 0.087854 -nan(ind) 1.47 [ 30% ] 3 0.087399 0.087838 -nan(ind) 1.48 [ 40% ] 4 0.087506 0.087854 -nan(ind) 1.47 [ 50% ] 5 0.087429 0.087842 -nan(ind) 1.48 [ 60% ] 6 0.087253 0.087858 -nan(ind) 1.47 [ 70% ] 7 0.087302 0.087832 -nan(ind) 1.47 [ 80% ] 8 0.087207 0.087886 -nan(ind) 1.47 [ 90% ] 9 0.087445 0.087895 -nan(ind) 1.49 [ 100% ] 10 0.087482 0.087847 -nan(ind) 1.48

我使用的数据格式如下： 0,0:0:59,1:1:1,2:21:1,3:2276:1,3:2632:1,3:2797:1 0,0:0:65,1:1:1,2:21:1,3:642:1,3:90:1 0,0:0:71,1:1:1,2:21:1,3:2013:1,3:2046:1 0,0:0:78,1:2:2,2:21:1,3:2381:1 0,0:0:56,1:1:1,2:21:1,3:1767:1,3:2628:1,3:2863:1,3:3209:1,3:43:1 0,0:0:62,1:1:1,2:21:1,3:1823:1,3:2228:1,3:3067:1 0,0:0:74,1:2:2,2:21:1,3:2972:1 0,0:0:79,1:1:1,2:21:1,3:1060:1,3:1268:1,3:1469:1,3:1481:1,3:1767:1,3:1820:1,3:1952:1,3:2276:1,3:2374:1,3:2994:1,3:3172:1,3:3852:1,3:4398:1,3:4462:1,3:819:1 0,0:0:68,1:1:1,2:21:1,3:1071:1,3:1072:1,3:43:1 0,0:0:34,1:2:2,2:21:1,3:43:1,3:4569:1 0,0:0:57,1:1:1,2:21:1,3:1781:1,3:3177:1,3:3191:1 0,0:0:68,1:1:1,2:21:1,3:3753:1 0,0:0:53,1:1:1,2:21:1,3:3439:1 0,0:0:64,1:2:2,2:21:1,3:2850:1,3:3012:1,3:3067:1,3:4016:1,3:4017:1,3:4188:1,3:668:1 1,0:0:76,1:1:1,2:21:1,3:632:1 0,0:0:82,1:2:2,2:21:1,3:2943:1,3:3482:1,3:3935:1,3:576:1 请问这是什么问题导致的？