wav2vec2-base-finetuned-ks

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1376
• Accuracy: 0.8210
• F1: 0.8209

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.3731	0.99	35	1.3532	0.3767	0.2859
1.3039	2.0	71	1.2740	0.4237	0.3434
1.2185	2.99	106	1.1573	0.5020	0.4423
1.0887	4.0	142	1.1107	0.5013	0.4389
1.0183	4.99	177	1.0801	0.5610	0.5348
0.8625	6.0	213	0.9364	0.6373	0.6285
0.7487	6.99	248	0.9735	0.6048	0.5867
0.6151	8.0	284	0.8946	0.6698	0.6735
0.5081	8.99	319	0.8748	0.6797	0.6855
0.4559	10.0	355	0.8701	0.6850	0.6832
0.4347	10.99	390	0.8887	0.7003	0.7040
0.2845	12.0	426	0.8715	0.7129	0.7145
0.275	12.99	461	0.8846	0.7268	0.7263
0.2301	14.0	497	0.8651	0.7261	0.7324
0.1657	14.99	532	0.8573	0.7473	0.7473
0.1593	16.0	568	0.8472	0.7420	0.7443
0.1398	16.99	603	0.7433	0.7825	0.7829
0.1318	18.0	639	0.7989	0.7739	0.7768
0.1425	18.99	674	0.7967	0.7759	0.7788
0.1116	20.0	710	0.8969	0.7659	0.7650
0.0716	20.99	745	0.9783	0.7434	0.7480
0.0909	22.0	781	0.9413	0.7593	0.7626
0.0691	22.99	816	0.9298	0.7832	0.7832
0.068	24.0	852	0.9522	0.7725	0.7744
0.0416	24.99	887	0.9624	0.7686	0.7746
0.0569	26.0	923	0.9376	0.7832	0.7832
0.0369	26.99	958	1.0163	0.7845	0.7843
0.0482	28.0	994	1.0013	0.7931	0.7895
0.0497	28.99	1029	1.1005	0.7725	0.7713
0.0427	30.0	1065	1.0346	0.7891	0.7901
0.0252	30.99	1100	1.0611	0.7871	0.7883
0.0268	32.0	1136	1.0436	0.7944	0.7962
0.022	32.99	1171	1.0217	0.8031	0.8012
0.0127	34.0	1207	1.0936	0.7971	0.7969
0.0153	34.99	1242	1.0777	0.8097	0.8055
0.0062	36.0	1278	1.2379	0.7699	0.7751
0.0081	36.99	1313	1.0697	0.7977	0.7987
0.0072	38.0	1349	1.1284	0.7997	0.8001
0.0105	38.99	1384	1.0593	0.8137	0.8136
0.0102	40.0	1420	1.0805	0.8130	0.8126
0.0088	40.99	1455	1.1237	0.8110	0.8115
0.0073	42.0	1491	1.0980	0.8170	0.8167
0.0046	42.99	1526	1.1584	0.8044	0.8049
0.0061	44.0	1562	1.1517	0.8110	0.8114
0.0021	44.99	1597	1.1564	0.8064	0.8074
0.0073	46.0	1633	1.1214	0.8183	0.8183
0.002	46.99	1668	1.1376	0.8210	0.8209
0.0064	48.0	1704	1.1283	0.8210	0.8208
0.0072	48.99	1739	1.1271	0.8203	0.8201
0.0019	49.3	1750	1.1273	0.8203	0.8201

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.1+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0