pretrained_qa_model

This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 4.6367

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: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 300
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
7.7719	16.6667	50	3.3236
2.8253	33.3333	100	3.3940
1.9882	50.0	150	3.9482
1.4762	66.6667	200	3.6320
1.4392	83.3333	250	6.2485
1.2776	100.0	300	4.3183
1.251	116.6667	350	4.1987
1.2782	133.3333	400	4.5603
1.0958	150.0	450	8.2640
1.0485	166.6667	500	5.5495
1.054	183.3333	550	5.3635
1.1684	200.0	600	1.8302
1.109	216.6667	650	6.1931
1.1607	233.3333	700	3.2514
1.0009	250.0	750	3.3236
1.2045	266.6667	800	10.1146
1.1297	283.3333	850	6.5903
0.9679	300.0	900	4.6367

Framework versions

• Transformers 4.49.0.dev0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0