mms-1b-all-swagen-baseline-62

This model is a fine-tuned version of facebook/mms-1b-all on the SWAGEN - SWA dataset. It achieves the following results on the evaluation set:

• Loss: 0.2190
• Wer: 0.1841

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: 8
• eval_batch_size: 4
• seed: 62
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.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: 100
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
8.0965	0.4773	100	4.5194	0.9996
3.8958	0.9547	200	3.7104	0.9984
3.5252	1.4296	300	3.3793	1.0035
3.2568	1.9069	400	3.1019	0.9986
3.0391	2.3819	500	3.0241	0.9998
2.7792	2.8592	600	0.6605	0.3460
0.3335	3.3341	700	0.2654	0.1989
0.2718	3.8115	800	0.2517	0.1886
0.2518	4.2864	900	0.2534	0.1925
0.2546	4.7637	1000	0.2488	0.1952
0.251	5.2387	1100	0.2456	0.1884
0.241	5.7160	1200	0.2401	0.1906
0.2281	6.1909	1300	0.2382	0.1878
0.2305	6.6683	1400	0.2327	0.1886
0.224	7.1432	1500	0.2294	0.1876
0.2224	7.6205	1600	0.2272	0.1851
0.2339	8.0955	1700	0.2295	0.1880
0.2181	8.5728	1800	0.2229	0.1874
0.2234	9.0477	1900	0.2257	0.1851
0.2184	9.5251	2000	0.2238	0.1861
0.2206	10.0	2100	0.2216	0.1845
0.2158	10.4773	2200	0.2206	0.1837
0.2134	10.9547	2300	0.2216	0.1837
0.2048	11.4296	2400	0.2189	0.1841
0.2143	11.9069	2500	0.2210	0.1851
0.2184	12.3819	2600	0.2214	0.1853
0.2046	12.8592	2700	0.2209	0.1839

Framework versions

• Transformers 4.53.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.0