xlm-roberta-base_original_kin-amh-eng_train_spearman_corr

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: 0.0298
• Spearman Corr: 0.7549

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: 2e-05
• 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	0.0259	0.7039
0.0435	3.51	400	0.0246	0.7125
0.0203	5.26	600	0.0272	0.7445
0.0142	7.02	800	0.0322	0.7419
0.0112	8.77	1000	0.0286	0.7447
0.0091	10.53	1200	0.0293	0.7442
0.0074	12.28	1400	0.0313	0.7552
0.0064	14.04	1600	0.0284	0.7563
0.0064	15.79	1800	0.0304	0.7570
0.0053	17.54	2000	0.0309	0.7540
0.0046	19.3	2200	0.0301	0.7557
0.0042	21.05	2400	0.0305	0.7558
0.0039	22.81	2600	0.0286	0.7569
0.0036	24.56	2800	0.0285	0.7591
0.0034	26.32	3000	0.0282	0.7575
0.0033	28.07	3200	0.0291	0.7538
0.0033	29.82	3400	0.0298	0.7549

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2