xlm-roberta-base_lr5e-06_seed42_basic_original_esp-kin-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: 0.0293
• Spearman Corr: 0.7251

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: 5e-06
• 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	1.63	200	0.0280	0.6016
0.062	3.27	400	0.0275	0.6391
0.0304	4.9	600	0.0293	0.6641
0.0245	6.53	800	0.0266	0.6888
0.0217	8.16	1000	0.0304	0.6900
0.0217	9.8	1200	0.0286	0.7016
0.0198	11.43	1400	0.0304	0.7080
0.0181	13.06	1600	0.0277	0.7103
0.0164	14.69	1800	0.0285	0.7086
0.0154	16.33	2000	0.0286	0.7233
0.0154	17.96	2200	0.0259	0.7209
0.0144	19.59	2400	0.0282	0.7160
0.0137	21.22	2600	0.0300	0.7168
0.0129	22.86	2800	0.0300	0.7215
0.0123	24.49	3000	0.0288	0.7262
0.0124	26.12	3200	0.0285	0.7256
0.0124	27.76	3400	0.0291	0.7220
0.0119	29.39	3600	0.0293	0.7251

Framework versions

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