saattrupdan/xlmr-base-texas-squad-fr

TExAS-SQuAD-fr

This model is a fine-tuned version of xlm-roberta-base on the TExAS-SQuAD-fr dataset. It achieves the following results on the evaluation set:

• Exact match: xx.xx%
• F1-score: xx.xx%

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• 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
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
2.1478	0.23	1000	1.8543
1.9827	0.46	2000	1.7643
1.8427	0.69	3000	1.6789
1.8372	0.92	4000	1.6137
1.7318	1.15	5000	1.6093
1.6603	1.38	6000	1.7157
1.6334	1.61	7000	1.6302
1.6716	1.84	8000	1.5845
1.5192	2.06	9000	1.6690
1.5174	2.29	10000	1.6669
1.4611	2.52	11000	1.6301
1.4648	2.75	12000	1.6009
1.5052	2.98	13000	1.6133

Framework versions

• Transformers 4.12.2
• Pytorch 1.8.1+cu101
• Datasets 1.12.1
• Tokenizers 0.10.3