SciBERT_25K_steps_bs64

This model is a fine-tuned version of allenai/scibert_scivocab_uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0177
• Accuracy: 0.9941
• Precision: 0.7990
• Recall: 0.5288
• F1: 0.6364
• Hamming: 0.0059

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 25000

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Hamming
0.0467	0.16	5000	0.0416	0.9902	0.0	0.0	0.0	0.0098
0.0236	0.32	10000	0.0223	0.9932	0.8192	0.3929	0.5311	0.0068
0.0198	0.47	15000	0.0190	0.9939	0.8015	0.4934	0.6108	0.0061
0.0185	0.63	20000	0.0180	0.9940	0.7974	0.5220	0.6310	0.0060
0.0181	0.79	25000	0.0177	0.9941	0.7990	0.5288	0.6364	0.0059

Framework versions

• Transformers 4.35.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.7.1
• Tokenizers 0.14.1