wav2vec2-large-mms-1b-kazakh-speech2ner-kscsyn-8b-4ep

This model is a fine-tuned version of facebook/mms-1b-all on the None 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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.6358	0.07	2000	6.5080	1.0000
6.6338	0.15	4000	6.5080	1.0000
0.0	0.22	6000	nan	1.0
0.0	0.3	8000	nan	1.0
0.0	0.37	10000	nan	1.0
0.0	0.44	12000	nan	1.0
0.0	0.52	14000	nan	1.0
0.0	0.59	16000	nan	1.0
0.0	0.66	18000	nan	1.0
0.0	0.74	20000	nan	1.0
0.0	0.81	22000	nan	1.0
0.0	0.89	24000	nan	1.0
0.0	0.96	26000	nan	1.0
0.0	1.03	28000	nan	1.0
0.0	1.11	30000	nan	1.0
0.0	1.18	32000	nan	1.0
0.0	1.25	34000	nan	1.0
0.0	1.33	36000	nan	1.0
0.0	1.4	38000	nan	1.0
0.0	1.48	40000	nan	1.0
0.0	1.55	42000	nan	1.0
0.0	1.62	44000	nan	1.0
0.0	1.7	46000	nan	1.0
0.0	1.77	48000	nan	1.0
0.0	1.84	50000	nan	1.0
0.0	1.92	52000	nan	1.0
0.0	1.99	54000	nan	1.0
0.0	2.07	56000	nan	1.0
0.0	2.14	58000	nan	1.0
0.0	2.21	60000	nan	1.0
0.0	2.29	62000	nan	1.0
0.0	2.36	64000	nan	1.0
0.0	2.43	66000	nan	1.0
0.0	2.51	68000	nan	1.0
0.0	2.58	70000	nan	1.0
0.0	2.66	72000	nan	1.0
0.0	2.73	74000	nan	1.0
0.0	2.8	76000	nan	1.0
0.0	2.88	78000	nan	1.0
0.0	2.95	80000	nan	1.0
0.0	3.02	82000	nan	1.0
0.0	3.1	84000	nan	1.0
0.0	3.17	86000	nan	1.0
0.0	3.25	88000	nan	1.0
0.0	3.32	90000	nan	1.0
0.0	3.39	92000	nan	1.0
0.0	3.47	94000	nan	1.0
0.0	3.54	96000	nan	1.0
0.0	3.61	98000	nan	1.0
0.0	3.69	100000	nan	1.0
0.0	3.76	102000	nan	1.0
0.0	3.84	104000	nan	1.0
0.0	3.91	106000	nan	1.0
0.0	3.98	108000	nan	1.0

Framework versions

• Transformers 4.33.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3