hubert_base_arabic_mdd

This model is a fine-tuned version of facebook/hubert-base-ls960 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.2265
• Wer: 1.0
• Per: 1.0

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Per
6.526	1.0	1637	3.3650	1.0	1.0
3.2555	2.0	3274	3.2755	1.0	1.0
3.2548	3.0	4911	3.2238	1.0	1.0
3.2385	4.0	6548	3.2845	1.0	1.0
3.2358	5.0	8185	3.2271	1.0	1.0
3.237	6.0	9822	3.2473	1.0	1.0
3.2622	7.0	11459	3.2289	1.0	1.0
3.2614	8.0	13096	3.2283	1.0	1.0
3.224	9.0	14733	3.2249	1.0	1.0
3.2221	10.0	16370	3.2335	1.0	1.0
3.222	11.0	18007	3.2357	1.0	1.0
3.2218	12.0	19644	3.2491	1.0	1.0
3.2183	13.0	21281	3.2446	1.0	1.0
3.2181	14.0	22918	3.2416	1.0	1.0
3.2164	15.0	24555	3.2259	1.0	1.0
3.2148	16.0	26192	3.2249	1.0	1.0
3.2139	17.0	27829	3.2327	1.0	1.0
3.2133	18.0	29466	3.2251	1.0	1.0
3.2128	19.0	31103	3.2288	1.0	1.0
3.2113	20.0	32740	3.2265	1.0	1.0

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0