llm_science_bert

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

• Loss: 1.6014
• Accuracy: 0.33

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: 0.003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	25	1.6123	0.24
No log	2.0	50	1.6094	0.235
1.6308	3.0	75	1.6111	0.245
1.6308	4.0	100	1.6090	0.26
1.6308	5.0	125	1.6099	0.275
1.625	6.0	150	1.6088	0.28
1.625	7.0	175	1.6092	0.285
1.625	8.0	200	1.6070	0.28
1.616	9.0	225	1.6059	0.27
1.616	10.0	250	1.6049	0.285
1.616	11.0	275	1.6051	0.3
1.6134	12.0	300	1.6052	0.3
1.6134	13.0	325	1.6048	0.305
1.6134	14.0	350	1.6038	0.315
1.6179	15.0	375	1.6042	0.3
1.6179	16.0	400	1.6040	0.315
1.6179	17.0	425	1.6035	0.32
1.6113	18.0	450	1.6031	0.325
1.6113	19.0	475	1.6030	0.32
1.6113	20.0	500	1.6024	0.305
1.631	21.0	525	1.6024	0.31
1.631	22.0	550	1.6024	0.315
1.631	23.0	575	1.6023	0.32
1.6153	24.0	600	1.6021	0.315
1.6153	25.0	625	1.6020	0.325
1.6153	26.0	650	1.6018	0.33
1.6091	27.0	675	1.6017	0.33
1.6091	28.0	700	1.6016	0.33
1.6091	29.0	725	1.6015	0.33
1.6125	30.0	750	1.6014	0.33

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.0
• Datasets 2.14.5
• Tokenizers 0.13.3