xtreme_s_xlsr_minds14

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the GOOGLE/XTREME_S - MINDS14 dataset. It achieves the following results on the evaluation set:

Loss: 0.2890
F1: 0.9474
Accuracy: 0.9470

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: 0.0003
train_batch_size: 32
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 2
total_train_batch_size: 64
total_eval_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 1500
num_epochs: 50.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Accuracy
2.551	2.7	200	2.5855	0.0407	0.1201
1.6934	5.41	400	1.5072	0.5862	0.6085
0.5914	8.11	600	0.7274	0.8270	0.8232
0.3896	10.81	800	0.4402	0.8905	0.8890
0.5052	13.51	1000	0.4483	0.8837	0.8829
0.4806	16.22	1200	0.4981	0.8784	0.8787
0.2103	18.92	1400	0.4957	0.8810	0.8817
0.4198	21.62	1600	0.5161	0.8927	0.8921
0.11	24.32	1800	0.4456	0.8923	0.8902
0.1233	27.03	2000	0.3858	0.9016	0.9012
0.1827	29.73	2200	0.3765	0.9162	0.9159
0.1235	32.43	2400	0.3716	0.9134	0.9128
0.1873	35.14	2600	0.3080	0.9314	0.9311
0.017	37.84	2800	0.2629	0.9415	0.9409
0.0436	40.54	3000	0.3159	0.9397	0.9390
0.0455	43.24	3200	0.2963	0.9393	0.9390
0.046	45.95	3400	0.2914	0.9457	0.9451
0.0042	48.65	3600	0.2890	0.9474	0.9470

Framework versions

Transformers 4.18.0.dev0
Pytorch 1.10.2+cu113
Datasets 1.18.4.dev0
Tokenizers 0.11.6