csml_word2vec_2

This model is a fine-tuned version of facebook/wav2vec2-base on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.1216
• Wer: 0.0732

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 500
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.3929	0.1778	100	3.1750	1.0
2.961	0.3556	200	2.9510	1.0
2.9448	0.5333	300	2.9355	1.0
2.9296	0.7111	400	2.9174	1.0
2.9131	0.8889	500	2.9153	1.0
2.9061	1.0658	600	2.9083	1.0
2.8503	1.2436	700	2.7161	1.0
0.9469	1.4213	800	0.7322	0.5705
0.5739	1.5991	900	0.3999	0.3134
0.4307	1.7769	1000	0.3026	0.2371
0.3733	1.9547	1100	0.2624	0.1979
0.2743	2.1316	1200	0.2269	0.1660
0.2504	2.3093	1300	0.2041	0.1504
0.2397	2.4871	1400	0.1879	0.1355
0.2652	2.6649	1500	0.1792	0.1327
0.2329	2.8427	1600	0.1692	0.1232
0.2049	3.0196	1700	0.1552	0.1111
0.1767	3.1973	1800	0.1528	0.1113
0.1521	3.3751	1900	0.1516	0.1049
0.1672	3.5529	2000	0.1461	0.1003
0.1704	3.7307	2100	0.1460	0.0996
0.1574	3.9084	2200	0.1456	0.0975
0.1293	4.0853	2300	0.1419	0.0933
0.1336	4.2631	2400	0.1350	0.0896
0.1562	4.4409	2500	0.1352	0.0870
0.1461	4.6187	2600	0.1394	0.0897
0.1359	4.7964	2700	0.1330	0.0860
0.141	4.9742	2800	0.1262	0.0835
0.1056	5.1511	2900	0.1294	0.0815
0.128	5.3289	3000	0.1261	0.0797
0.1433	5.5067	3100	0.1248	0.0808
0.1129	5.6844	3200	0.1264	0.0783
0.1245	5.8622	3300	0.1263	0.0790
0.1215	6.0391	3400	0.1264	0.0772
0.097	6.2169	3500	0.1222	0.0755
0.0955	6.3947	3600	0.1243	0.0748
0.1026	6.5724	3700	0.1238	0.0750
0.1061	6.7502	3800	0.1219	0.0732
0.1031	6.928	3900	0.1230	0.0737
0.0959	7.1049	4000	0.1216	0.0732

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1