xlm-roberta-base_original_kin-hau-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: 0.0372
• Spearman Corr: 0.7228

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	1.59	200	0.0438	0.6734
0.0472	3.19	400	0.0432	0.7134
0.0246	4.78	600	0.0376	0.7083
0.0185	6.37	800	0.0347	0.7099
0.0143	7.97	1000	0.0435	0.7061
0.0143	9.56	1200	0.0379	0.7193
0.0111	11.16	1400	0.0392	0.7231
0.0093	12.75	1600	0.0388	0.7170
0.0078	14.34	1800	0.0403	0.7242
0.0067	15.94	2000	0.0383	0.7229
0.0067	17.53	2200	0.0395	0.7170
0.0059	19.12	2400	0.0377	0.7133
0.005	20.72	2600	0.0359	0.7181
0.0047	22.31	2800	0.0354	0.7211
0.0045	23.9	3000	0.0361	0.7161
0.0045	25.5	3200	0.0357	0.7202
0.0041	27.09	3400	0.0372	0.7228

Framework versions

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