distilbert-fa-zwnj-base-finetuned-pquad

This model is a fine-tuned version of HooshvareLab/distilbert-fa-zwnj-base on the pquad dataset.

Results

Test set

name	value
'exact'	66.38340414896275,
'f1'	80.23760220987583,
'total'	8002,
'HasAns_exact'	60.13469119579501,
'HasAns_f1'	78.34449620292781,
'HasAns_total'	6088,
'NoAns_exact'	86.25914315569489,
'NoAns_f1'	86.25914315569489,
'NoAns_total'	1914,
'best_exact'	66.38340414896275,
'best_exact_thresh'	0.0,
'best_f1'	80.23760220987589,
'best_f1_thresh'	0.0

Validation set

name	value
'exact'	64.65646940822468,
'f1'	78.88641788270802,
'total'	7976,
'HasAns_exact'	57.54795663052544,
'HasAns_f1'	76.4800782372771,
'HasAns_total'	5995,
'NoAns_exact'	86.16860171630489,
'NoAns_f1'	86.16860171630489,
'NoAns_total'	1981,
'best_exact'	64.65646940822468,
'best_exact_thresh'	0.0,
'best_f1'	78.88641788270819,
'best_f1_thresh'	0.0

Model description

uses distilbert-fa-zwnj-base as base and fine-tunes it on pquad dataset.

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: 16
• seed: 42
• 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
1.1299	1.0	4003	1.1306
0.845	2.0	8006	1.0839
0.639	3.0	12009	1.1302

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.9.0
• Tokenizers 0.13.2