wav2vec2-base-finetuned-ie

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: 0.9461
• Accuracy: 0.6857

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1913	1.0	102	1.1911	0.4374
1.0238	2.0	204	1.0045	0.5529
0.809	3.0	306	0.9396	0.6188
0.7097	4.0	408	1.2146	0.5558
0.6082	5.0	510	1.2860	0.5752
0.4099	6.0	612	1.3618	0.5771
0.3927	7.0	714	1.1155	0.6508
0.2013	8.0	816	1.3554	0.6266
0.2208	9.0	918	1.7306	0.5674
0.1967	10.0	1020	1.6680	0.6004
0.1563	11.0	1122	1.6125	0.6402
0.0777	12.0	1224	1.7766	0.6305
0.0486	13.0	1326	1.8744	0.6324
0.0594	14.0	1428	1.9529	0.6246
0.0363	15.0	1530	1.8843	0.6334

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3