wav2vec2-large-960h-lv60-self-intent-classification-ori

This model is a fine-tuned version of facebook/wav2vec2-large-960h-lv60-self on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1985
• Accuracy: 0.5417

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
2.2033	1.0	14	2.2126	0.0833
2.2006	2.0	28	2.2026	0.0833
2.1786	3.0	42	2.1758	0.3333
2.1712	4.0	56	2.1436	0.3333
2.1495	5.0	70	2.1120	0.3333
2.1326	6.0	84	2.0909	0.3333
2.1039	7.0	98	2.0966	0.3333
2.0931	8.0	112	2.0355	0.3333
2.1144	9.0	126	2.0082	0.3333
2.0258	10.0	140	1.9901	0.375
2.0028	11.0	154	1.9429	0.3958
1.9737	12.0	168	1.9538	0.3958
1.9023	13.0	182	1.8824	0.375
1.9226	14.0	196	1.8607	0.3958
1.8521	15.0	210	1.8065	0.3958
1.7752	16.0	224	1.8153	0.4167
1.8391	17.0	238	1.7470	0.4375
1.7041	18.0	252	1.7419	0.4167
1.7075	19.0	266	1.6644	0.4375
1.6845	20.0	280	1.6340	0.4375
1.6275	21.0	294	1.6271	0.4167
1.4586	22.0	308	1.5640	0.4375
1.4987	23.0	322	1.5279	0.4583
1.5513	24.0	336	1.4873	0.4792
1.4828	25.0	350	1.4887	0.4583
1.4711	26.0	364	1.4613	0.4583
1.371	27.0	378	1.4062	0.4792
1.3789	28.0	392	1.4038	0.4792
1.3579	29.0	406	1.4031	0.4792
1.2771	30.0	420	1.3637	0.5
1.3417	31.0	434	1.3655	0.5
1.231	32.0	448	1.3698	0.5
1.2367	33.0	462	1.3394	0.5
1.2933	34.0	476	1.3448	0.4792
1.1631	35.0	490	1.2867	0.5417
1.165	36.0	504	1.2624	0.5417
1.2431	37.0	518	1.2252	0.5625
1.1731	38.0	532	1.2082	0.5625
1.1734	39.0	546	1.2062	0.5417
1.1631	40.0	560	1.2034	0.5417
1.0963	41.0	574	1.1973	0.5417
1.2157	42.0	588	1.1988	0.5625
1.1467	43.0	602	1.2018	0.5417
1.1503	44.0	616	1.1986	0.5417
1.0945	45.0	630	1.1985	0.5417

Framework versions

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