scibert-lm-v1-finetuned-20

This model is a fine-tuned version of allenai/scibert_scivocab_cased on the conll2003 dataset. It achieves the following results on the evaluation set:

• Loss: 22.6145

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.0118	1.0	1756	15.0609
0.0001	2.0	3512	17.9265
0.0	3.0	5268	18.6256
0.0001	4.0	7024	19.5144
0.0002	5.0	8780	19.8926
0.0	6.0	10536	21.6975
0.0	7.0	12292	22.2388
0.0	8.0	14048	21.0441
0.0	9.0	15804	21.6852
0.0	10.0	17560	22.4439
0.0	11.0	19316	20.9994
0.0	12.0	21072	21.7275
0.0	13.0	22828	22.1329
0.0	14.0	24584	22.4599
0.0	15.0	26340	22.5726
0.0	16.0	28096	22.7823
0.0	17.0	29852	22.4167
0.0	18.0	31608	22.4075
0.0	19.0	33364	22.5731
0.0	20.0	35120	22.6145

Framework versions

• Transformers 4.21.1
• Pytorch 1.12.0+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1