Edit model card

ai-light-dance_singing4_ft_wav2vec2-large-xlsr-53-5gram-v4-2

This model is a fine-tuned version of gary109/ai-light-dance_singing2_ft_wav2vec2-large-xlsr-53-5gram-v4-2 on the GARY109/AI_LIGHT_DANCE - ONSET-SINGING4 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2311
  • Wer: 0.1042

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 4e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 100.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.9329 1.0 72 0.4334 0.1349
0.7631 2.0 144 0.3390 0.1318
0.7575 3.0 216 0.3182 0.1441
0.667 4.0 288 0.2995 0.1288
0.6474 5.0 360 0.3211 0.1574
0.6094 6.0 432 0.2944 0.1530
0.5586 7.0 504 0.3446 0.1809
0.5728 8.0 576 0.2682 0.1246
0.575 9.0 648 0.2610 0.1244
0.5882 10.0 720 0.2463 0.1199
0.5367 11.0 792 0.2542 0.1093
0.5261 12.0 864 0.2523 0.1131
0.5091 13.0 936 0.2491 0.1175
0.5323 14.0 1008 0.2535 0.1208
0.5478 15.0 1080 0.2542 0.1204
0.4775 16.0 1152 0.2476 0.1138
0.4922 17.0 1224 0.2535 0.1162
0.4893 18.0 1296 0.2501 0.1143
0.4672 19.0 1368 0.2703 0.1162
0.4764 20.0 1440 0.2632 0.1253
0.4716 21.0 1512 0.2534 0.1097
0.4733 22.0 1584 0.2496 0.1086
0.4577 23.0 1656 0.2637 0.1225
0.4714 24.0 1728 0.2489 0.1102
0.4615 25.0 1800 0.2399 0.1020
0.4636 26.0 1872 0.2311 0.1042
0.4564 27.0 1944 0.2413 0.1042
0.4452 28.0 2016 0.2436 0.1037
0.4336 29.0 2088 0.2484 0.1070
0.4628 30.0 2160 0.2385 0.1005
0.4475 31.0 2232 0.2446 0.1075
0.4264 32.0 2304 0.2548 0.1068
0.4417 33.0 2376 0.2442 0.1046
0.4165 34.0 2448 0.2458 0.1045
0.4398 35.0 2520 0.2475 0.0979
0.4334 36.0 2592 0.2375 0.1027
0.4279 37.0 2664 0.2462 0.1053
0.4213 38.0 2736 0.2402 0.1028
0.4394 39.0 2808 0.2385 0.1020
0.4415 40.0 2880 0.2428 0.1021
0.4173 41.0 2952 0.2356 0.1016
0.4006 42.0 3024 0.2461 0.1007
0.4055 43.0 3096 0.2412 0.0998
0.4163 44.0 3168 0.2378 0.0976
0.4143 45.0 3240 0.2471 0.0996
0.4132 46.0 3312 0.2457 0.1004
0.3991 47.0 3384 0.2350 0.1019
0.4014 48.0 3456 0.2400 0.1025
0.416 49.0 3528 0.2370 0.1010
0.4067 50.0 3600 0.2444 0.1010
0.3876 51.0 3672 0.2491 0.1057
0.3964 52.0 3744 0.2451 0.1075
0.3903 53.0 3816 0.2395 0.1003
0.4036 54.0 3888 0.2446 0.1016
0.3936 55.0 3960 0.2520 0.0997
0.4094 56.0 4032 0.2401 0.0992
0.3977 57.0 4104 0.2498 0.1019
0.3942 58.0 4176 0.2496 0.0989
0.4052 59.0 4248 0.2507 0.1021
0.3995 60.0 4320 0.2382 0.0999
0.407 61.0 4392 0.2517 0.1037
0.4067 62.0 4464 0.2430 0.1034
0.3887 63.0 4536 0.2415 0.0974
0.3837 64.0 4608 0.2435 0.0991
0.3954 65.0 4680 0.2384 0.0985
0.3726 66.0 4752 0.2550 0.1009
0.3659 67.0 4824 0.2523 0.0967
0.376 68.0 4896 0.2571 0.0973
0.3759 69.0 4968 0.2528 0.0981
0.3862 70.0 5040 0.2496 0.0976
0.367 71.0 5112 0.2465 0.0942
0.3688 72.0 5184 0.2505 0.0968
0.3817 73.0 5256 0.2525 0.0973
0.3675 74.0 5328 0.2441 0.0964
0.3727 75.0 5400 0.2440 0.0973
0.371 76.0 5472 0.2510 0.0971
0.3761 77.0 5544 0.2398 0.0961
0.358 78.0 5616 0.2485 0.0956
0.3521 79.0 5688 0.2438 0.0955
0.3722 80.0 5760 0.2422 0.0967
0.3663 81.0 5832 0.2463 0.0949
0.3716 82.0 5904 0.2467 0.0965
0.361 83.0 5976 0.2458 0.0957
0.3626 84.0 6048 0.2501 0.0942
0.3755 85.0 6120 0.2486 0.0943
0.3693 86.0 6192 0.2497 0.0947
0.3548 87.0 6264 0.2515 0.0958
0.3583 88.0 6336 0.2433 0.0937
0.3678 89.0 6408 0.2456 0.0933
0.3718 90.0 6480 0.2443 0.0937
0.3614 91.0 6552 0.2457 0.0957
0.3541 92.0 6624 0.2450 0.0953
0.3671 93.0 6696 0.2469 0.0944
0.3503 94.0 6768 0.2459 0.0956
0.3692 95.0 6840 0.2461 0.0944
0.362 96.0 6912 0.2430 0.0945
0.3431 97.0 6984 0.2454 0.0952
0.3597 98.0 7056 0.2454 0.0940
0.354 99.0 7128 0.2448 0.0939
0.3597 100.0 7200 0.2449 0.0943

Framework versions

  • Transformers 4.21.0.dev0
  • Pytorch 1.9.1+cu102
  • Datasets 2.3.3.dev0
  • Tokenizers 0.12.1
Downloads last month
20
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.