apac_5sents_XLS-R_2_1e-6

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 267.7123
  • Wer: 1.0

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: 1e-06
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • training_steps: 5000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
277.1571 5.54 100 267.0546 1.0
276.4313 11.11 200 261.9242 1.0
265.9415 16.65 300 250.8011 1.0
238.6939 22.22 400 203.5241 1.0
174.4503 27.76 500 144.1427 1.0
133.3724 33.32 600 115.6477 1.0
111.8372 38.86 700 101.5777 1.0
100.9788 44.43 800 93.3811 1.0
93.1826 49.97 900 88.1119 1.0
88.9464 55.54 1000 84.3092 1.0
85.3244 61.11 1100 81.3484 1.0
81.8194 66.65 1200 78.9506 1.0
80.4364 72.22 1300 76.9131 1.0
77.8947 77.76 1400 75.1301 1.0
76.6954 83.32 1500 73.5633 1.0
74.5509 88.86 1600 72.1455 1.0
73.6445 94.43 1700 70.8411 1.0
72.1876 99.97 1800 69.6471 1.0
71.3292 105.54 1900 68.5407 1.0
70.0705 111.11 2000 67.4980 1.0
68.7978 116.65 2100 66.5158 1.0
68.0348 122.22 2200 65.5796 1.0
66.8713 127.76 2300 64.6976 1.0
66.16 133.32 2400 63.8526 1.0
65.2028 138.86 2500 63.0523 1.0
64.6085 144.43 2600 62.2911 1.0
63.5373 149.97 2700 61.5736 1.0
63.2075 155.54 2800 60.8822 1.0
62.4467 161.11 2900 60.2360 1.0
61.4208 166.65 3000 59.6212 1.0
61.1933 172.22 3100 59.0409 1.0
60.4399 177.76 3200 58.4939 1.0
60.0438 183.32 3300 57.9786 1.0
59.1202 188.86 3400 57.4928 1.0
59.1382 194.43 3500 57.0438 1.0
58.3211 199.97 3600 56.6216 1.0
58.1737 205.54 3700 56.2324 1.0
57.8397 211.11 3800 55.8701 1.0
57.1469 216.65 3900 55.5402 1.0
57.0738 222.22 4000 55.2410 1.0
56.6554 227.76 4100 54.9684 1.0
56.6029 233.32 4200 54.7281 1.0
55.9671 238.86 4300 54.5120 1.0
56.0056 244.43 4400 54.3272 1.0
55.7506 249.97 4500 54.1668 1.0
55.7896 255.54 4600 54.0354 1.0
55.6986 261.11 4700 53.9389 1.0
55.3473 266.65 4800 53.8633 1.0
55.4997 272.22 4900 53.8202 1.0
55.2924 277.76 5000 53.8053 1.0

Framework versions

  • Transformers 4.26.1
  • Pytorch 2.0.1+cu118
  • Datasets 2.11.0
  • Tokenizers 0.13.3
Downloads last month
145
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Evaluation results