xlm-roberta-quality-classifier

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Accuracy: 0.9815
• F1: 0.9817
• F1 High: 0.9741
• F1 Low: 0.9989
• F1 Medium: 0.9720
• Loss: 0.0651
• Precision: 0.9818
• Recall: 0.9816

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: 2.0000000000000003e-06
• train_batch_size: 96
• eval_batch_size: 256
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 192
• total_eval_batch_size: 512
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Accuracy	F1	F1 High	F1 Low	F1 Medium	Validation Loss	Precision	Recall
0.036	3.7736	1000	0.9849	0.9850	0.9799	0.9973	0.9779	0.0388	0.9852	0.9853
0.0149	7.5472	2000	0.9815	0.9817	0.9741	0.9989	0.9720	0.0651	0.9818	0.9816

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1