delirium_roberta

This model is a fine-tuned version of allenai/biomed_roberta_base. It achieves the following results on the evaluation set:

• Loss: 0.3709

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: 0.0005
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• lr_scheduler_warmup_steps: 100
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss
1.2088	0.4	100	0.8023
0.8075	0.8	200	0.7029
0.7404	1.2	300	0.6575
0.6826	1.6	400	0.6096
0.6578	2.0	500	0.5995
0.6525	2.4	600	0.5834
0.6223	2.8	700	0.5650
0.6	3.2	800	0.5464
0.5807	3.6	900	0.5312
0.5963	4.0	1000	0.5233
0.584	4.4	1100	0.5154
0.5508	4.8	1200	0.5049
0.5609	5.2	1300	0.4960
0.5397	5.6	1400	0.4851
0.5401	6.0	1500	0.4805
0.513	6.4	1600	0.4690
0.5247	6.8	1700	0.4647
0.5228	7.2	1800	0.4607
0.5142	7.6	1900	0.4534
0.5055	8.0	2000	0.4428
0.4942	8.4	2100	0.4338
0.4895	8.8	2200	0.4336
0.4874	9.2	2300	0.4221
0.4744	9.6	2400	0.4234
0.4743	10.0	2500	0.4139
0.4816	10.4	2600	0.4090
0.4733	10.8	2700	0.4077
0.4419	11.2	2800	0.4035
0.4552	11.6	2900	0.3989
0.4467	12.0	3000	0.3913
0.45	12.4	3100	0.3884
0.4551	12.8	3200	0.3864
0.4247	13.2	3300	0.3786
0.4432	13.6	3400	0.3874
0.4086	14.0	3500	0.3776
0.4308	14.4	3600	0.3711
0.4293	14.8	3700	0.3763
0.4235	15.2	3800	0.3733
0.4138	15.6	3900	0.3758
0.4156	16.0	4000	0.3709

Framework versions

• Transformers 4.34.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1