This model is a fine-tuned version of ylacombe/w2v-bert-2.0 on the MOZILLA-FOUNDATION/COMMON_VOICE_16_0 - TR dataset. It achieves the following results on the evaluation set:

Loss: nan
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: 0.007448827845832091
train_batch_size: 20
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 40
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 5000
num_epochs: 15.0
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.27	300	3.2930	1.0
5.6462	0.55	600	3.4159	1.0
5.6462	0.82	900	3.4422	1.0
3.3522	1.1	1200	3.3719	1.0
3.2605	1.37	1500	3.4026	1.0
3.2605	1.64	1800	3.4448	1.0
3.2766	1.92	2100	3.4736	0.9999
3.2766	2.19	2400	3.9828	1.0
3.2853	2.47	2700	3.5532	1.0
3.3389	2.74	3000	3.7819	1.0
3.3389	3.01	3300	3.2250	1.0
3.2186	3.29	3600	3.2373	1.0
3.2186	3.56	3900	3.2162	1.0
3.1916	3.84	4200	3.2368	1.0
3.2188	4.11	4500	3.2377	1.0
3.2188	4.38	4800	3.4207	1.0
5.3067	4.66	5100	nan	1.0
5.3067	4.93	5400	nan	1.0
0.0	5.21	5700	nan	1.0
0.0	5.48	6000	nan	1.0
0.0	5.75	6300	nan	1.0
0.0	6.03	6600	nan	1.0
0.0	6.3	6900	nan	1.0
0.0	6.58	7200	nan	1.0
0.0	6.85	7500	nan	1.0
0.0	7.12	7800	nan	1.0
0.0	7.4	8100	nan	1.0
0.0	7.67	8400	nan	1.0
0.0	7.95	8700	nan	1.0
0.0	8.22	9000	nan	1.0
0.0	8.49	9300	nan	1.0
0.0	8.77	9600	nan	1.0
0.0	9.04	9900	nan	1.0
0.0	9.32	10200	nan	1.0
0.0	9.59	10500	nan	1.0
0.0	9.86	10800	nan	1.0
0.0	10.14	11100	nan	1.0
0.0	10.41	11400	nan	1.0
0.0	10.68	11700	nan	1.0
0.0	10.96	12000	nan	1.0
0.0	11.23	12300	nan	1.0
0.0	11.51	12600	nan	1.0
0.0	11.78	12900	nan	1.0
0.0	12.05	13200	nan	1.0
0.0	12.33	13500	nan	1.0
0.0	12.6	13800	nan	1.0
0.0	12.88	14100	nan	1.0
0.0	13.15	14400	nan	1.0
0.0	13.42	14700	nan	1.0
0.0	13.7	15000	nan	1.0
0.0	13.97	15300	nan	1.0
0.0	14.25	15600	nan	1.0
0.0	14.52	15900	nan	1.0
0.0	14.79	16200	nan	1.0

Framework versions

Transformers 4.37.0.dev0
Pytorch 2.1.0+cu121
Datasets 2.14.5
Tokenizers 0.15.0

ylacombe/wav2vec2-common_voice-tr-demo

wav2vec2-common_voice-tr-demo

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

ylacombe/
wav2vec2-common_voice-tr-demo