ewc_stabilised

This model is a fine-tuned version of masakhane/afroxlmr-large-ner-masakhaner-1.0_2.0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1396
• F1: 0.8317
• Precision: 0.8305
• Recall: 0.8328
• Accuracy: 0.9605

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: 16
• eval_batch_size: 8
• seed: 3407
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Precision	Recall	Accuracy
0.3184	0.9993	701	0.1480	0.7895	0.7950	0.7841	0.9511
0.1333	2.0	1403	0.1271	0.8195	0.8148	0.8242	0.9578
0.0975	2.9993	2104	0.1241	0.8289	0.8254	0.8324	0.9598
0.0744	4.0	2806	0.1293	0.8307	0.8313	0.8300	0.9603
0.0596	4.9964	3505	0.1396	0.8317	0.8305	0.8328	0.9605

Framework versions

• Transformers 4.43.4
• Pytorch 2.4.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1