BERT_BASE_TS_phonetic_wikitext_0.01

This model is a fine-tuned version of bert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7949

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: 0.0004
• train_batch_size: 256
• eval_batch_size: 256
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-06
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10000
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.9703	0.3019	2000	1.8355
2.05	0.6039	4000	1.9240
2.1918	0.9058	6000	2.0171
2.3388	1.2077	8000	2.1917
2.5037	1.5097	10000	2.3411
2.5405	1.8116	12000	2.3671
2.5213	2.1135	14000	2.3876
2.5101	2.4155	16000	2.3514
2.48	2.7174	18000	2.3582
2.4458	3.0193	20000	2.3223
2.4225	3.3213	22000	2.2794
2.3855	3.6232	24000	2.2808
2.3565	3.9251	26000	2.2409
2.3209	4.2271	28000	2.1957
2.3006	4.5290	30000	2.1681
2.2685	4.8309	32000	2.2009
2.2217	5.1329	34000	2.1588
2.1991	5.4348	36000	2.0944
2.1697	5.7367	38000	2.0508
2.1464	6.0386	40000	2.0425
2.1115	6.3406	42000	2.0055
2.0749	6.6425	44000	2.0070
2.0685	6.9444	46000	1.9856
2.0253	7.2464	48000	1.9362
1.9794	7.5483	50000	1.9261
1.9784	7.8502	52000	1.9403
1.9359	8.1522	54000	1.8783
1.9237	8.4541	56000	1.8597
1.8984	8.7560	58000	1.8374
1.869	9.0580	60000	1.8380
1.8506	9.3599	62000	1.8130
1.84	9.6618	64000	1.7785
1.8374	9.9638	66000	1.7949

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0