fine-tuned-DatasetQAS-TYDI-QA-ID-with-xlm-roberta-large-with-ITTL-without-freeze-LR-1e-05

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

• Loss: 0.9402
• Exact Match: 69.3662
• F1: 82.0036

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match	F1
6.2837	0.5	19	3.6986	8.4507	17.7536
6.2837	0.99	38	2.5899	18.4859	29.7766
3.6833	1.5	57	1.7044	42.6056	56.8157
3.6833	1.99	76	1.2711	53.3451	70.2979
3.6833	2.5	95	1.1063	62.3239	75.7765
1.5024	2.99	114	1.0275	64.2606	78.0460
1.5024	3.5	133	0.9941	65.8451	79.1313
1.0028	3.99	152	0.9642	67.4296	80.6196
1.0028	4.5	171	0.9682	69.0141	82.4975
1.0028	4.99	190	0.9455	67.9577	81.0386
0.7765	5.5	209	0.9802	67.7817	81.0844
0.7765	5.99	228	0.9402	69.3662	82.0036

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu117
• Datasets 2.2.0
• Tokenizers 0.13.2