google-bert-large-uncased-arabic-fp16-allagree

This model is a fine-tuned version of google-bert/bert-large-uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3863
• Accuracy: 0.8554
• Precision: 0.8580
• Recall: 0.8554
• F1: 0.8563

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: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.3
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
1.0847	0.7463	50	0.8073	0.6838	0.7469	0.6838	0.6079
0.7434	1.4925	100	0.5989	0.7631	0.7790	0.7631	0.7325
0.5675	2.2388	150	0.4732	0.8181	0.8114	0.8181	0.8125
0.5108	2.9851	200	0.4486	0.8162	0.8107	0.8162	0.8123
0.4261	3.7313	250	0.4340	0.8368	0.8427	0.8368	0.8304
0.3764	4.4776	300	0.3863	0.8554	0.8580	0.8554	0.8563
0.3405	5.2239	350	0.3998	0.8573	0.8566	0.8573	0.8535
0.2931	5.9701	400	0.4547	0.8330	0.8533	0.8330	0.8378
0.243	6.7164	450	0.4321	0.8451	0.8567	0.8451	0.8489

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1