Hugging Face's logo

cdactvm
/
w2v_bert_malayalam_100125

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2-bert
Generated from Trainer
Inference Endpoints
Model card Files Community
w2v_bert_malayalam_100125 / README.md
cdactvm's picture
cdactvm
End of training
c9d1222 verified
preview code
|
raw
history blame contribute delete
13 kB
metadata
library_name: transformers
license: mit
base_model: facebook/w2v-bert-2.0
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: w2v_bert_malayalam_100125
    results: []

w2v_bert_malayalam_100125

This model is a fine-tuned version of facebook/w2v-bert-2.0 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2922
  • Wer: 0.2847

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: 5e-05
  • train_batch_size: 2
  • eval_batch_size: 1
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 8
  • optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 5
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
13.7406 0.0270 300 1.1859 0.8493
6.3024 0.0539 600 0.9225 0.6676
5.3208 0.0809 900 0.7547 0.6066
5.0544 0.1078 1200 0.7201 0.6010
4.552 0.1348 1500 0.6276 0.5619
4.0527 0.1617 1800 0.6411 0.5327
3.8443 0.1887 2100 0.6421 0.5384
3.7674 0.2156 2400 0.6330 0.5094
3.7184 0.2426 2700 0.5730 0.5052
3.567 0.2696 3000 0.5459 0.4992
3.633 0.2965 3300 0.5636 0.4906
3.473 0.3235 3600 0.5271 0.4986
3.3079 0.3504 3900 0.5346 0.4924
3.6181 0.3774 4200 0.5180 0.4706
3.3054 0.4043 4500 0.5266 0.4706
3.2021 0.4313 4800 0.5171 0.4714
3.237 0.4583 5100 0.4940 0.4442
3.2646 0.4852 5400 0.5097 0.4691
3.2935 0.5122 5700 0.5393 0.4513
3.2023 0.5391 6000 0.4942 0.4491
3.1896 0.5661 6300 0.4971 0.4706
3.169 0.5930 6600 0.4829 0.4387
3.1288 0.6200 6900 0.4699 0.4299
3.0414 0.6469 7200 0.4697 0.4202
3.1903 0.6739 7500 0.4547 0.4260
3.1618 0.7009 7800 0.4512 0.4496
2.8182 0.7278 8100 0.4548 0.4369
2.9933 0.7548 8400 0.4716 0.4215
2.8348 0.7817 8700 0.4551 0.4380
2.9557 0.8087 9000 0.4702 0.4429
2.9472 0.8356 9300 0.4360 0.4405
2.9094 0.8626 9600 0.4693 0.4278
3.0442 0.8895 9900 0.4419 0.4198
2.7861 0.9165 10200 0.4344 0.4079
2.9604 0.9435 10500 0.4338 0.4125
2.8824 0.9704 10800 0.4474 0.4136
2.9609 0.9974 11100 0.4267 0.4129
2.8458 1.0243 11400 0.4365 0.3914
2.9807 1.0512 11700 0.4217 0.3915
2.7052 1.0782 12000 0.4294 0.3914
2.7261 1.1051 12300 0.4059 0.3979
2.757 1.1321 12600 0.4055 0.3924
2.6792 1.1590 12900 0.4168 0.3943
2.6238 1.1860 13200 0.4132 0.4120
2.704 1.2130 13500 0.4135 0.3984
2.7817 1.2399 13800 0.4213 0.3953
2.7019 1.2669 14100 0.4136 0.3903
2.683 1.2938 14400 0.4008 0.3857
2.6341 1.3208 14700 0.3932 0.3921
2.8109 1.3477 15000 0.3920 0.3886
2.7188 1.3747 15300 0.4107 0.3775
2.787 1.4016 15600 0.4017 0.3832
2.5539 1.4286 15900 0.3919 0.3870
2.7399 1.4556 16200 0.4052 0.3787
2.6653 1.4825 16500 0.3994 0.3781
2.6334 1.5095 16800 0.3962 0.3804
2.6855 1.5364 17100 0.3904 0.3799
2.5878 1.5634 17400 0.3874 0.3738
2.6821 1.5903 17700 0.4038 0.3801
2.7367 1.6173 18000 0.3895 0.3786
2.5238 1.6442 18300 0.3802 0.3714
2.6262 1.6712 18600 0.3871 0.3735
2.6882 1.6982 18900 0.3718 0.3598
2.6244 1.7251 19200 0.3690 0.3702
2.5328 1.7521 19500 0.3749 0.3696
2.7317 1.7790 19800 0.3849 0.3671
2.7712 1.8060 20100 0.3799 0.3572
2.5236 1.8329 20400 0.3669 0.3586
2.5933 1.8599 20700 0.3695 0.3699
2.6017 1.8869 21000 0.3794 0.3608
2.6945 1.9138 21300 0.3683 0.3660
2.4709 1.9408 21600 0.3681 0.3566
2.3483 1.9677 21900 0.3668 0.3583
2.441 1.9947 22200 0.3765 0.3623
2.3229 2.0216 22500 0.3814 0.3570
2.4638 2.0485 22800 0.3653 0.3535
2.4375 2.0755 23100 0.3715 0.3556
2.449 2.1024 23400 0.3664 0.3539
2.3533 2.1294 23700 0.3648 0.3522
2.5918 2.1563 24000 0.3697 0.3495
2.2601 2.1833 24300 0.3645 0.3509
2.4091 2.2103 24600 0.3633 0.3481
2.5612 2.2372 24900 0.3947 0.3475
2.4217 2.2642 25200 0.3683 0.3538
2.4534 2.2911 25500 0.3564 0.3521
2.4084 2.3181 25800 0.3620 0.3489
2.3584 2.3450 26100 0.3761 0.3561
2.2511 2.3720 26400 0.3603 0.3495
2.4207 2.3989 26700 0.3563 0.3455
2.4695 2.4259 27000 0.3571 0.3428
2.6855 2.4529 27300 0.3468 0.3471
2.3552 2.4798 27600 0.3503 0.3436
2.3278 2.5068 27900 0.3561 0.3503
2.3505 2.5337 28200 0.3532 0.3504
2.472 2.5607 28500 0.3460 0.3463
2.3524 2.5876 28800 0.3551 0.3483
2.4979 2.6146 29100 0.3512 0.3322
2.3248 2.6416 29400 0.3572 0.3491
2.5329 2.6685 29700 0.3395 0.3474
2.4015 2.6955 30000 0.3545 0.3382
2.3657 2.7224 30300 0.3484 0.3422
2.3756 2.7494 30600 0.3436 0.3396
2.4377 2.7763 30900 0.3462 0.3300
2.4235 2.8033 31200 0.3405 0.3319
2.4171 2.8302 31500 0.3743 0.3426
2.2713 2.8572 31800 0.3443 0.3285
2.3465 2.8842 32100 0.3480 0.3441
2.2693 2.9111 32400 0.3538 0.3374
2.2837 2.9381 32700 0.3352 0.3316
2.2519 2.9650 33000 0.3453 0.3425
2.3385 2.9920 33300 0.3369 0.3328
2.4399 3.0189 33600 0.3369 0.3314
2.1657 3.0458 33900 0.3354 0.3210
2.1836 3.0728 34200 0.3418 0.3305
2.1411 3.0997 34500 0.3403 0.3274
2.1968 3.1267 34800 0.3431 0.3271
2.1438 3.1536 35100 0.3344 0.3203
2.2291 3.1806 35400 0.3370 0.3304
2.2565 3.2076 35700 0.3379 0.3211
2.2529 3.2345 36000 0.3323 0.3172
2.1685 3.2615 36300 0.3289 0.3204
2.0921 3.2884 36600 0.3380 0.3371
2.2647 3.3154 36900 0.3278 0.3212
2.1798 3.3423 37200 0.3404 0.3267
2.0501 3.3693 37500 0.3318 0.3171
2.1228 3.3963 37800 0.3377 0.3117
2.2038 3.4232 38100 0.3312 0.3161
2.113 3.4502 38400 0.3170 0.3131
2.3311 3.4771 38700 0.3291 0.3179
2.1042 3.5041 39000 0.3219 0.3159
2.2017 3.5310 39300 0.3449 0.3168
2.1555 3.5580 39600 0.3239 0.3091
2.0275 3.5849 39900 0.3214 0.3108
2.1272 3.6119 40200 0.3313 0.3141
2.1742 3.6389 40500 0.3145 0.3104
2.2524 3.6658 40800 0.3098 0.3073
2.3791 3.6928 41100 0.3129 0.3151
2.1903 3.7197 41400 0.3140 0.3086
2.1773 3.7467 41700 0.3170 0.3122
2.2465 3.7736 42000 0.3137 0.3113
2.152 3.8006 42300 0.3090 0.3050
2.0966 3.8275 42600 0.3133 0.3034
2.0236 3.8545 42900 0.3065 0.3053
2.2719 3.8815 43200 0.3177 0.3038
2.0735 3.9084 43500 0.3057 0.3036
2.0077 3.9354 43800 0.3083 0.2995
2.2148 3.9623 44100 0.3100 0.3061
1.9275 3.9893 44400 0.3193 0.3001
2.0617 4.0162 44700 0.3018 0.3014
1.97 4.0431 45000 0.2992 0.3017
2.0957 4.0701 45300 0.3084 0.3047
2.0003 4.0970 45600 0.3127 0.2997
2.0239 4.1240 45900 0.3080 0.2988
1.8299 4.1510 46200 0.3096 0.2993
2.0207 4.1779 46500 0.3116 0.2990
2.3016 4.2049 46800 0.2990 0.2965
2.0119 4.2318 47100 0.2991 0.2978
2.0965 4.2588 47400 0.3046 0.2969
2.0322 4.2857 47700 0.2995 0.2982
1.8958 4.3127 48000 0.3045 0.2984
2.0243 4.3396 48300 0.3046 0.2936
2.0465 4.3666 48600 0.3049 0.2937
1.9224 4.3936 48900 0.2986 0.2910
2.0303 4.4205 49200 0.3027 0.2925
1.9259 4.4475 49500 0.3035 0.2931
2.1682 4.4744 49800 0.3020 0.2921
1.9361 4.5014 50100 0.2984 0.2924
1.9593 4.5283 50400 0.2980 0.2887
2.0082 4.5553 50700 0.2959 0.2878
2.0995 4.5822 51000 0.2977 0.2886
1.9609 4.6092 51300 0.2950 0.2892
1.8096 4.6362 51600 0.2979 0.2890
1.8145 4.6631 51900 0.2978 0.2877
1.8261 4.6901 52200 0.2958 0.2871
1.8683 4.7170 52500 0.2939 0.2854
2.0299 4.7440 52800 0.2899 0.2851
2.0949 4.7709 53100 0.2916 0.2848
1.8456 4.7979 53400 0.2911 0.2854
1.9542 4.8249 53700 0.2932 0.2837
1.8429 4.8518 54000 0.2942 0.2866
1.9042 4.8788 54300 0.2939 0.2852
2.0831 4.9057 54600 0.2903 0.2848
1.8793 4.9327 54900 0.2912 0.2851
1.7786 4.9596 55200 0.2917 0.2850
1.9494 4.9866 55500 0.2922 0.2847

Framework versions

  • Transformers 4.47.1
  • Pytorch 2.5.1+cu124
  • Datasets 3.2.0
  • Tokenizers 0.21.0