xlm-roberta-base_lr0.0001_seed42_basic_original_kin-amh-eng_train

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0399
• Spearman Corr: 0.0444

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: 32
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Spearman Corr
No log	1.75	200	0.0384	-0.0298
0.0606	3.51	400	0.0428	0.0555
0.0542	5.26	600	0.0399	0.1268
0.0522	7.02	800	0.0373	0.2132
0.0519	8.77	1000	0.0399	0.0780
0.0516	10.53	1200	0.0391	nan
0.051	12.28	1400	0.0380	-0.0558
0.0505	14.04	1600	0.0375	0.0632
0.0505	15.79	1800	0.0390	0.0344
0.0505	17.54	2000	0.0395	0.0575
0.0498	19.3	2200	0.0397	0.0553
0.0494	21.05	2400	0.0381	0.0810
0.0492	22.81	2600	0.0383	nan
0.0489	24.56	2800	0.0399	0.0444

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2