gpt2-NER-favsbot

This model is a fine-tuned version of gpt2 on the favsbot dataset. It achieves the following results on the evaluation set:

• Loss: 1.5146
• Precision: 0.3782
• Recall: 0.3278
• F1: 0.3512
• Accuracy: 0.5597

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: 1.5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	4	4.0808	0.0630	0.0444	0.0521	0.0773
No log	2.0	8	3.5205	0.0946	0.0778	0.0854	0.1077
No log	3.0	12	3.0413	0.0909	0.0722	0.0805	0.2084
No log	4.0	16	2.6817	0.0615	0.0444	0.0516	0.2740
No log	5.0	20	2.4227	0.1140	0.0722	0.0884	0.3560
No log	6.0	24	2.2422	0.1382	0.0944	0.1122	0.3770
No log	7.0	28	2.0941	0.1654	0.1222	0.1406	0.3864
No log	8.0	32	1.9726	0.2344	0.1667	0.1948	0.4309
No log	9.0	36	1.8916	0.2925	0.1722	0.2168	0.4543
No log	10.0	40	1.8321	0.31	0.1722	0.2214	0.4660
No log	11.0	44	1.7697	0.2957	0.1889	0.2305	0.4707
No log	12.0	48	1.7087	0.3228	0.2278	0.2671	0.4965
No log	13.0	52	1.6551	0.3485	0.2556	0.2949	0.5152
No log	14.0	56	1.6136	0.3219	0.2611	0.2883	0.5176
No log	15.0	60	1.5819	0.3510	0.2944	0.3202	0.5363
No log	16.0	64	1.5575	0.3506	0.3	0.3234	0.5410
No log	17.0	68	1.5394	0.3529	0.3	0.3243	0.5433
No log	18.0	72	1.5265	0.3791	0.3222	0.3483	0.5574
No log	19.0	76	1.5180	0.3766	0.3222	0.3473	0.5574
No log	20.0	80	1.5146	0.3782	0.3278	0.3512	0.5597

Framework versions

• Transformers 4.21.1
• Pytorch 1.12.1
• Datasets 2.4.0
• Tokenizers 0.12.1