Edit model card

QA_FineTuned_Arabert

This model is a fine-tuned version of zohaib99k/Bert_Arabic-SQuADv2-QA on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 3.3343

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: 0.0002
  • train_batch_size: 12
  • eval_batch_size: 12
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 48
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 5

Training results

Training Loss Epoch Step Validation Loss
0.1603 0.46 10 2.7965
0.3055 0.92 20 3.0322
0.3091 1.38 30 2.8940
0.2692 1.84 40 2.9621
0.2198 2.3 50 2.9107
0.2112 2.76 60 3.1322
0.1576 3.22 70 3.2085
0.1392 3.68 80 3.1323
0.1474 4.14 90 3.2893
0.0905 4.6 100 3.3343

Framework versions

  • Transformers 4.39.3
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.15.2
Downloads last month
42
Safetensors
Model size
135M params
Tensor type
F32
ยท

Finetuned from