AmedeoBonatti/nlp_te_class_scibert_tok

class_scibert_nopretrain

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

• Loss: 1.1004
• Accuracy: 0.8933

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: 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: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	56	0.3029	0.88
No log	2.0	112	0.3203	0.8844
No log	3.0	168	0.4238	0.9067
No log	4.0	224	0.5721	0.8667
No log	5.0	280	0.6623	0.88
No log	6.0	336	0.6685	0.8933
No log	7.0	392	0.6864	0.9067
No log	8.0	448	0.7105	0.8978
0.0902	9.0	504	0.7304	0.8978
0.0902	10.0	560	0.7451	0.8978
0.0902	11.0	616	0.7603	0.8978
0.0902	12.0	672	0.7714	0.8978
0.0902	13.0	728	0.7845	0.8978
0.0902	14.0	784	0.7961	0.8978
0.0902	15.0	840	0.8059	0.8978
0.0902	16.0	896	0.8171	0.8978
0.0902	17.0	952	0.8258	0.8978
0.0001	18.0	1008	0.8337	0.8978
0.0001	19.0	1064	0.8420	0.8978
0.0001	20.0	1120	0.8510	0.8978
0.0001	21.0	1176	0.8590	0.8978
0.0001	22.0	1232	0.8681	0.8978
0.0001	23.0	1288	0.8738	0.8978
0.0001	24.0	1344	0.8802	0.8978
0.0001	25.0	1400	0.8858	0.8933
0.0001	26.0	1456	0.8928	0.8933
0.0	27.0	1512	0.9000	0.8933
0.0	28.0	1568	0.9057	0.8933
0.0	29.0	1624	0.9097	0.8933
0.0	30.0	1680	0.9170	0.8933
0.0	31.0	1736	0.9224	0.8933
0.0	32.0	1792	0.9259	0.8933
0.0	33.0	1848	0.9303	0.8933
0.0	34.0	1904	0.9361	0.8933
0.0	35.0	1960	0.9418	0.8933
0.0	36.0	2016	0.9477	0.8933
0.0	37.0	2072	0.9521	0.8933
0.0	38.0	2128	0.9558	0.8933
0.0	39.0	2184	0.9597	0.8933
0.0	40.0	2240	0.9645	0.8933
0.0	41.0	2296	0.9681	0.8933
0.0	42.0	2352	0.9722	0.8933
0.0	43.0	2408	0.9759	0.8933
0.0	44.0	2464	0.9797	0.8933
0.0	45.0	2520	0.9836	0.8933
0.0	46.0	2576	0.9874	0.8933
0.0	47.0	2632	0.9919	0.8933
0.0	48.0	2688	0.9952	0.8933
0.0	49.0	2744	0.9993	0.8933
0.0	50.0	2800	1.0033	0.8933
0.0	51.0	2856	1.0066	0.8933
0.0	52.0	2912	1.0109	0.8933
0.0	53.0	2968	1.0144	0.8933
0.0	54.0	3024	1.0177	0.8933
0.0	55.0	3080	1.0210	0.8933
0.0	56.0	3136	1.0242	0.8933
0.0	57.0	3192	1.0272	0.8933
0.0	58.0	3248	1.0315	0.8933
0.0	59.0	3304	1.0340	0.8933
0.0	60.0	3360	1.0364	0.8933
0.0	61.0	3416	1.0392	0.8933
0.0	62.0	3472	1.0423	0.8933
0.0	63.0	3528	1.0452	0.8933
0.0	64.0	3584	1.0475	0.8933
0.0	65.0	3640	1.0504	0.8933
0.0	66.0	3696	1.0529	0.8933
0.0	67.0	3752	1.0543	0.8933
0.0	68.0	3808	1.0578	0.8933
0.0	69.0	3864	1.0606	0.8933
0.0	70.0	3920	1.0629	0.8933
0.0	71.0	3976	1.0650	0.8933
0.0	72.0	4032	1.0666	0.8933
0.0	73.0	4088	1.0690	0.8933
0.0	74.0	4144	1.0712	0.8933
0.0	75.0	4200	1.0731	0.8933
0.0	76.0	4256	1.0751	0.8933
0.0	77.0	4312	1.0766	0.8933
0.0	78.0	4368	1.0784	0.8933
0.0	79.0	4424	1.0805	0.8933
0.0	80.0	4480	1.0822	0.8933
0.0	81.0	4536	1.0836	0.8933
0.0	82.0	4592	1.0860	0.8933
0.0	83.0	4648	1.0873	0.8933
0.0	84.0	4704	1.0888	0.8933
0.0	85.0	4760	1.0902	0.8933
0.0	86.0	4816	1.0916	0.8933
0.0	87.0	4872	1.0927	0.8933
0.0	88.0	4928	1.0937	0.8933
0.0	89.0	4984	1.0948	0.8933
0.0	90.0	5040	1.0959	0.8933
0.0	91.0	5096	1.0965	0.8933
0.0	92.0	5152	1.0972	0.8933
0.0	93.0	5208	1.0982	0.8933
0.0	94.0	5264	1.0988	0.8933
0.0	95.0	5320	1.0993	0.8933
0.0	96.0	5376	1.0998	0.8933
0.0	97.0	5432	1.1000	0.8933
0.0	98.0	5488	1.1001	0.8933
0.0	99.0	5544	1.1003	0.8933
0.0	100.0	5600	1.1004	0.8933

Framework versions

• Transformers 4.41.2
• Pytorch 2.2.1
• Datasets 2.19.2
• Tokenizers 0.19.1