mms-MGB2

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: 14
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
6.8035	0.01	250	5.9077	1.0036
2.2196	0.02	500	2.0224	0.9764
1.0641	0.03	750	0.8949	0.4840
0.8089	0.04	1000	0.7188	0.4095
1.7071	0.05	1250	0.7008	0.3974
0.8132	0.06	1500	0.6975	0.3986
0.9741	0.07	1750	0.6975	0.3986
0.8332	0.08	2000	0.6975	0.3986
0.8908	0.09	2250	0.6975	0.3986
0.8321	0.1	2500	0.6975	0.3986
0.7957	0.1	2750	0.6975	0.3986
0.9173	0.11	3000	0.6975	0.3986
2.0065	0.12	3250	0.6975	0.3986
0.8618	0.13	3500	0.6975	0.3986
0.9001	0.14	3750	0.6975	0.3986
1.0321	0.15	4000	0.6975	0.3986
0.8408	0.16	4250	0.6975	0.3986
0.8901	0.17	4500	0.6975	0.3986
0.8242	0.18	4750	0.6975	0.3986
0.8678	0.19	5000	0.6975	0.3986
0.8633	0.2	5250	0.6975	0.3986
0.8087	0.21	5500	0.6975	0.3986
0.9243	0.22	5750	0.6975	0.3986
0.7973	0.23	6000	0.6975	0.3986
0.835	0.24	6250	0.6975	0.3986
1.3251	0.25	6500	0.6975	0.3986
0.0	0.26	6750	nan	1.0
0.0	0.27	7000	nan	1.0
0.0	0.28	7250	nan	1.0
0.0	0.29	7500	nan	1.0
0.0	0.29	7750	nan	1.0
0.0	0.3	8000	nan	1.0
0.0	0.31	8250	nan	1.0
0.0	0.32	8500	nan	1.0
0.0	0.33	8750	nan	1.0
0.0	0.34	9000	nan	1.0
0.0	0.35	9250	nan	1.0
0.0	0.36	9500	nan	1.0
0.0	0.37	9750	nan	1.0
0.0	0.38	10000	nan	1.0

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1
• Datasets 2.19.1
• Tokenizers 0.13.3