wav2vec2-xlsr-53-espeak-cv-ft-intent-classification-ori

This model is a fine-tuned version of facebook/wav2vec2-xlsr-53-espeak-cv-ft on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.9124
• Accuracy: 0.625

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 45

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1894	1.0	14	2.1812	0.3333
2.1795	2.0	28	2.1553	0.3333
2.144	3.0	42	2.1066	0.3333
2.1175	4.0	56	2.0283	0.3542
2.0542	5.0	70	1.9253	0.3958
2.0007	6.0	84	1.8468	0.4167
1.8891	7.0	98	1.7655	0.4583
1.8484	8.0	112	1.6695	0.4792
1.8256	9.0	126	1.5920	0.5
1.6832	10.0	140	1.5331	0.5
1.6149	11.0	154	1.4763	0.5
1.5853	12.0	168	1.4453	0.5
1.4357	13.0	182	1.3588	0.5
1.4789	14.0	196	1.3238	0.4792
1.3886	15.0	210	1.2822	0.4792
1.313	16.0	224	1.2609	0.5
1.3559	17.0	238	1.2191	0.5208
1.1937	18.0	252	1.1936	0.5
1.1847	19.0	266	1.1547	0.5417
1.197	20.0	280	1.1390	0.5417
1.1057	21.0	294	1.1310	0.5208
1.0291	22.0	308	1.1086	0.5417
1.0768	23.0	322	1.1075	0.5417
1.0249	24.0	336	1.0654	0.5625
1.0433	25.0	350	1.0390	0.5625
0.9974	26.0	364	1.0086	0.6458
0.9578	27.0	378	0.9939	0.625
0.916	28.0	392	0.9938	0.625
0.9187	29.0	406	0.9843	0.625
0.8759	30.0	420	0.9755	0.625
0.9199	31.0	434	0.9822	0.6042
0.8791	32.0	448	0.9522	0.6458
0.8436	33.0	462	0.9414	0.6458
0.8692	34.0	476	0.9510	0.625
0.8201	35.0	490	0.9208	0.6667
0.8284	36.0	504	0.9398	0.6458
0.8761	37.0	518	0.9438	0.6458
0.7948	38.0	532	0.9253	0.6667
0.8339	39.0	546	0.9250	0.6458
0.8002	40.0	560	0.9145	0.6458
0.7791	41.0	574	0.9062	0.6667
0.7944	42.0	588	0.9077	0.6667
0.7777	43.0	602	0.9069	0.6458
0.7943	44.0	616	0.9118	0.625
0.7573	45.0	630	0.9124	0.625

Framework versions

• Transformers 4.20.1
• Pytorch 1.11.0
• Datasets 2.1.0
• Tokenizers 0.12.1