xlsr-nmcpc-nomi

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

Model description

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 0.0004
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
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: 132
num_epochs: 200
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
4.8042	6.0606	200	3.0647	0.9452
2.8665	12.1212	400	2.2960	0.9838
1.4496	18.1818	600	0.5882	0.6085
0.4807	24.2424	800	0.4014	0.4828
0.275	30.3030	1000	0.4216	0.3996
0.1757	36.3636	1200	0.2956	0.3651
0.1298	42.4242	1400	0.4517	0.3712
0.1018	48.4848	1600	0.4099	0.3529
0.08	54.5455	1800	0.3337	0.3651
0.0729	60.6061	2000	0.3765	0.3671
0.0604	66.6667	2200	0.3915	0.3671
0.0504	72.7273	2400	0.3723	0.3590
0.0449	78.7879	2600	0.3246	0.3489
0.0392	84.8485	2800	0.3044	0.3428
0.036	90.9091	3000	0.2869	0.3286
0.0424	96.9697	3200	0.3328	0.3408
0.0308	103.0303	3400	0.3950	0.3387
0.0296	109.0909	3600	0.3217	0.3306
0.0209	115.1515	3800	0.3163	0.3347
0.0179	121.2121	4000	0.3692	0.3387
0.0234	127.2727	4200	0.3597	0.3327
0.0141	133.3333	4400	0.3497	0.3266
0.0125	139.3939	4600	0.3291	0.3225
0.0125	145.4545	4800	0.3130	0.3185
0.0116	151.5152	5000	0.3337	0.3327
0.0117	157.5758	5200	0.3424	0.3367
0.0088	163.6364	5400	0.3385	0.3347
0.0087	169.6970	5600	0.3302	0.3266
0.0062	175.7576	5800	0.3093	0.3286
0.0064	181.8182	6000	0.3367	0.3286
0.0053	187.8788	6200	0.3370	0.3306
0.007	193.9394	6400	0.3354	0.3266
0.0058	200.0	6600	0.3286	0.3266

Safetensors

Model size

0.3B params

Tensor type

F32

Base model

Finetuned

(319)

this model