Metrics Upon Eval with max_length = 512

• loss: 2.4489
• accuracy: 0.7250
• precision: 0.7150
• recall: 0.7250
• precision_macro: 0.6583
• recall_macro: 0.6262
• macro_fpr: 0.0278
• weighted_fpr: 0.0264
• weighted_specificity: 0.9597
• macro_specificity: 0.9790
• weighted_sensitivity: 0.7250
• macro_sensitivity: 0.6262
• f1_micro: 0.7250
• f1_macro: 0.6317
• f1_weighted: 0.7155
• runtime: 27.7396
• samples_per_second: 46.5400
• steps_per_second: 5.8400

tiny-mistral

This model is a fine-tuned version of openaccess-ai-collective/tiny-mistral on an unknown dataset. It achieves the following results on the evaluation set (at last epoch):

• Loss: 2.5607
• Accuracy: 0.7126
• Precision: 0.7033
• Recall: 0.7126
• Precision Macro: 0.6443
• Recall Macro: 0.5942
• Macro Fpr: 0.0292
• Weighted Fpr: 0.0282
• Weighted Specificity: 0.9577
• Macro Specificity: 0.9779
• Weighted Sensitivity: 0.7111
• Macro Sensitivity: 0.5942
• F1 Micro: 0.7111
• F1 Macro: 0.6107
• F1 Weighted: 0.7086

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: 5e-05
• 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: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	Precision Macro	Recall Macro	Macro Fpr	Weighted Fpr	Weighted Specificity	Macro Specificity	Weighted Sensitivity	Macro Sensitivity	F1 Micro	F1 Macro	F1 Weighted
1.4479	1.0	643	1.1182	0.6499	0.6258	0.6499	0.4712	0.4744	0.0390	0.0371	0.9470	0.9731	0.6499	0.4744	0.6499	0.4547	0.6214
0.8133	2.0	1286	1.0854	0.6987	0.7197	0.6987	0.5877	0.5528	0.0305	0.0299	0.9608	0.9773	0.6987	0.5528	0.6987	0.5474	0.6970
0.5592	3.0	1929	1.6114	0.6987	0.7107	0.6987	0.6368	0.5881	0.0304	0.0299	0.9609	0.9773	0.6987	0.5881	0.6987	0.6013	0.6998
0.2375	4.0	2572	1.7779	0.6956	0.7001	0.6956	0.5840	0.5667	0.0310	0.0303	0.9566	0.9768	0.6956	0.5667	0.6956	0.5699	0.6923
0.1586	5.0	3215	2.1752	0.6948	0.7011	0.6948	0.5797	0.5799	0.0316	0.0304	0.9601	0.9770	0.6948	0.5799	0.6948	0.5695	0.6917
0.0956	6.0	3858	2.3261	0.7080	0.7213	0.7080	0.6169	0.6191	0.0291	0.0286	0.9646	0.9782	0.7080	0.6191	0.7080	0.6115	0.7105
0.044	7.0	4501	2.3308	0.7157	0.7143	0.7157	0.6184	0.5939	0.0285	0.0276	0.9611	0.9785	0.7157	0.5939	0.7157	0.6014	0.7131
0.0212	8.0	5144	2.5607	0.7126	0.7033	0.7126	0.6494	0.6175	0.0294	0.0280	0.9581	0.9780	0.7126	0.6175	0.7126	0.6237	0.7047
0.0183	9.0	5787	2.6405	0.7119	0.7092	0.7119	0.6133	0.5850	0.0291	0.0281	0.9599	0.9781	0.7119	0.5850	0.7119	0.5935	0.7088
0.0145	10.0	6430	2.7268	0.7088	0.7058	0.7088	0.6235	0.5945	0.0297	0.0285	0.9574	0.9777	0.7088	0.5945	0.7088	0.6039	0.7051
0.0065	11.0	7073	2.7568	0.7149	0.7133	0.7149	0.6342	0.5966	0.0286	0.0277	0.9609	0.9784	0.7149	0.5966	0.7149	0.6068	0.7123
0.0012	12.0	7716	2.9243	0.7088	0.7106	0.7088	0.6261	0.5886	0.0296	0.0285	0.9581	0.9778	0.7088	0.5886	0.7088	0.6011	0.7071
0.0019	13.0	8359	2.9101	0.7119	0.7107	0.7119	0.6399	0.5910	0.0291	0.0281	0.9576	0.9780	0.7119	0.5910	0.7119	0.6073	0.7085
0.0011	14.0	9002	2.9270	0.7103	0.7101	0.7103	0.6430	0.5925	0.0293	0.0283	0.9576	0.9779	0.7103	0.5925	0.7103	0.6090	0.7077
0.0008	15.0	9645	2.9390	0.7111	0.7110	0.7111	0.6443	0.5942	0.0292	0.0282	0.9577	0.9779	0.7111	0.5942	0.7111	0.6107	0.7086

Framework versions

• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2