Training my own dataset, but the indices equal 0 and in the output path (coco json format) don't have any caption.

[MingNarongthat]

I want to train my own dataset and built the data like VATEX, after I try to train this dataset I got the result from output pred.LS.public_test_32frames.beam1.max20_coco_format.json that there were no caption on this file. (comparing with the best checkpoint of VATEX) Anyone faced this issue and can help me for this situation? And this is my log.
07/20/2022 01:51:44 - INFO - main - ModelSaver save trial NO. 0
07/20/2022 01:51:56 - INFO - main - Save checkpoint to output/ls_default1/checkpoint-8-232
07/20/2022 01:51:56 - INFO - main - Perform evaluation at iteration 232, global_step 232
07/20/2022 01:51:57 - INFO - main - Inference model computing time: 0.05405998229980469 seconds per batch
07/20/2022 01:52:06 - INFO - main - evaluation result: {'Bleu_1': 0.0, 'Bleu_2': 0.0, 'Bleu_3': 0.0, 'Bleu_4': 0.0, 'METEOR': 0.0, 'ROUGE_L': 0.0, 'CIDEr': 0.0, 'SPICE': 0.0}
07/20/2022 01:52:06 - INFO - main - evaluation result saved to output/ls_default1/checkpoint-8-232/pred.LS.public_test_32frames.beam1.max20.eval.json
07/20/2022 01:52:08 - INFO - main - eta: 0:00:17 iter: 240 global_step: 240 speed: 16.4 images/sec loss: 4.3013 (5.2290) acc: 0.1379 (0.1169) batch_time: 0.3488 (0.3479) data_time: 0.0001 (0.0002) lr (Visual Encoder): 2.99e-06 lr (LM): 5.98e-05 max mem: 7632
07/20/2022 01:52:16 - INFO - main - eta: 0:00:10 iter: 260 global_step: 260 speed: 68.0 images/sec loss: 4.1436 (5.1596) acc: 0.1250 (0.1181) batch_time: 0.3514 (0.3483) data_time: 0.0001 (0.0002) lr (Visual Encoder): 1.84e-06 lr (LM): 3.68e-05 max mem: 7632
07/20/2022 01:52:16 - INFO - main - ModelSaver save trial NO. 0
07/20/2022 01:52:28 - INFO - main - Save checkpoint to output/ls_default1/checkpoint-9-261
07/20/2022 01:52:28 - INFO - main - Perform evaluation at iteration 261, global_step 261
07/20/2022 01:52:30 - INFO - main - Inference model computing time: 0.05387449264526367 seconds per batch
07/20/2022 01:52:38 - INFO - main - evaluation result: {'Bleu_1': 0.0, 'Bleu_2': 0.0, 'Bleu_3': 0.0, 'Bleu_4': 0.0, 'METEOR': 0.0, 'ROUGE_L': 0.0, 'CIDEr': 0.0, 'SPICE': 0.0}
07/20/2022 01:52:38 - INFO - main - evaluation result saved to output/ls_default1/checkpoint-9-261/pred.LS.public_test_32frames.beam1.max20.eval.json
07/20/2022 01:52:45 - INFO - main - eta: 0:00:03 iter: 280 global_step: 280 speed: 16.3 images/sec loss: 4.0845 (5.0965) acc: 0.1481 (0.1193) batch_time: 0.3515 (0.3487) data_time: 0.0001 (0.0002) lr (Visual Encoder): 6.90e-07 lr (LM): 1.38e-05 max mem: 7632
07/20/2022 01:52:49 - INFO - main - eta: 0:00:00 iter: 290 global_step: 290 speed: 136.6 images/sec loss: 4.1788 (5.0666) acc: 0.1600 (0.1207) batch_time: 0.3501 (0.3488) data_time: 0.0001 (0.0002) lr (Visual Encoder): 1.15e-07 lr (LM): 2.30e-06 max mem: 7632
07/20/2022 01:52:49 - INFO - main - ModelSaver save trial NO. 0
07/20/2022 01:53:02 - INFO - main - Save checkpoint to output/ls_default1/checkpoint-10-290
07/20/2022 01:53:02 - INFO - main - Perform evaluation at iteration 290, global_step 290
07/20/2022 01:53:03 - INFO - main - Inference model computing time: 0.05416369438171387 seconds per batch
07/20/2022 01:53:12 - INFO - main - evaluation result: {'Bleu_1': 0.0, 'Bleu_2': 0.0, 'Bleu_3': 0.0, 'Bleu_4': 0.0, 'METEOR': 0.0, 'ROUGE_L': 0.0, 'CIDEr': 0.0, 'SPICE': 0.0}
07/20/2022 01:53:12 - INFO - main - evaluation result saved to output/ls_default1/checkpoint-10-290/pred.LS.public_test_32frames.beam1.max20.eval.json
07/20/2022 01:53:12 - INFO - main - Total training time: 0:05:55.848689 (1.2271 s / iter)