xlm-roberta-large-xnli-v4.0

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

• Loss: 0.4963
• F1 Macro: 0.8192
• F1 Micro: 0.8204
• Accuracy Balanced: 0.8190
• Accuracy: 0.8204
• Precision Macro: 0.8193
• Recall Macro: 0.8190
• Precision Micro: 0.8204
• Recall Micro: 0.8204

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: 9e-06
• train_batch_size: 8
• eval_batch_size: 64
• seed: 40
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Macro	F1 Micro	Accuracy Balanced	Accuracy	Precision Macro	Recall Macro	Precision Micro	Recall Micro
0.3593	1.69	200	0.4297	0.8211	0.8218	0.8224	0.8218	0.8206	0.8224	0.8218	0.8218

eval result

Datasets	asadfgglie/nli-zh-tw-all/test	asadfgglie/BanBan_2024-10-17-facial_expressions-nli/test	eval_dataset	test_dataset
eval_loss	0.494	0.773	0.483	0.496
eval_f1_macro	0.821	0.627	0.825	0.819
eval_f1_micro	0.822	0.644	0.826	0.82
eval_accuracy_balanced	0.821	0.638	0.826	0.819
eval_accuracy	0.822	0.644	0.826	0.82
eval_precision_macro	0.821	0.663	0.825	0.819
eval_recall_macro	0.821	0.638	0.826	0.819
eval_precision_micro	0.822	0.644	0.826	0.82
eval_recall_micro	0.822	0.644	0.826	0.82
eval_runtime	50.82	0.635	10.346	39.781
eval_samples_per_second	167.257	1490.523	164.308	170.938
eval_steps_per_second	2.617	23.634	2.61	2.69
Size of dataset	8500	946	1700	6800

Framework versions

• Transformers 4.33.3
• Pytorch 2.5.1+cu121
• Datasets 2.14.7
• Tokenizers 0.13.3