SciBERT_AsymmetricLoss_25K_bs64_P1_N1

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: 67.0896
• Accuracy: 0.9945
• Precision: 0.7586
• Recall: 0.6438
• F1: 0.6965
• Hamming: 0.0055

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: 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
83.6475	0.16	5000	79.3653	0.9938	0.7361	0.5667	0.6404	0.0062
75.8712	0.32	10000	72.7250	0.9942	0.7513	0.6068	0.6714	0.0058
72.4202	0.47	15000	69.4174	0.9944	0.7568	0.6237	0.6838	0.0056
70.0693	0.63	20000	67.8098	0.9945	0.7561	0.6385	0.6923	0.0055
68.9765	0.79	25000	67.0896	0.9945	0.7586	0.6438	0.6965	0.0055

Framework versions

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