flanT5_MT

This model is a fine-tuned version of google/flan-t5-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9144
• Accuracy: 0.7959
• Precision: 0.8188
• Recall: 0.76
• F1 score: 0.7883

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: 0.0001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Accuracy	F1 score	Precision	Recall	Validation Loss
1.1594	0.2103	2500	0.7306	0.6914	0.8091	0.6035	1.1807
0.956	0.4205	5000	0.78	0.7556	0.85	0.68	1.0125
0.8973	0.6308	7500	1.2023	0.7529	0.8772	0.5882	0.7042
0.9154	0.8410	10000	1.0591	0.7771	0.8458	0.6776	0.7524
0.8148	1.0513	12500	1.1675	0.7753	0.8087	0.7212	0.7624
0.6499	1.2616	15000	0.9862	0.8076	0.8501	0.7471	0.7952
0.6059	1.4718	17500	1.0780	0.7659	0.7404	0.8188	0.7777
0.5391	1.6821	20000	1.2307	0.7694	0.7928	0.7294	0.7598
0.479	1.8923	22500	1.2428	0.7735	0.7675	0.7847	0.7760
0.3085	2.1026	25000	1.3597	0.7676	0.7571	0.7882	0.7723
0.226	2.3129	27500	1.6552	0.7776	0.7757	0.7812	0.7784
0.2293	2.5231	30000	1.4472	0.7847	0.7909	0.7741	0.7824
0.2201	2.7334	32500	1.3059	0.7982	0.7972	0.8	0.7986
0.2119	2.9437	35000	1.6964	0.7882	0.7981	0.7718	0.7847
0.087	3.1539	37500	1.9933	0.7818	0.7801	0.7847	0.7824
0.102	3.3642	40000	1.6337	0.7859	0.7866	0.7847	0.7856
0.0925	3.5744	42500	1.8106	0.7894	0.7808	0.8047	0.7926
0.1071	3.7847	45000	1.6925	0.7865	0.7691	0.8188	0.7932
0.077	3.9950	47500	1.8706	0.7929	0.8044	0.7741	0.7890
0.036	4.2052	50000	2.0159	0.7865	0.7822	0.7941	0.7881
0.0534	4.4155	52500	1.9290	0.7882	0.7862	0.7918	0.7890
0.0516	4.6257	55000	1.9351	0.7959	0.8180	0.7612	0.7885
0.0471	4.8360	57500	1.9144	0.7959	0.8188	0.76	0.7883

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1