MT_Complaint

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

• Loss: 0.4773

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.4907	1.0	151	0.3250
0.2887	2.0	302	0.2513
0.2371	3.0	453	0.2255
0.1945	4.0	604	0.2192
0.1683	5.0	755	0.2119
0.1461	6.0	906	0.2207
0.1143	7.0	1057	0.2182
0.1011	8.0	1208	0.2421
0.0818	9.0	1359	0.2476
0.0799	10.0	1510	0.2660
0.0757	11.0	1661	0.2661
0.0619	12.0	1812	0.2687
0.0517	13.0	1963	0.2939
0.0468	14.0	2114	0.3191
0.0352	15.0	2265	0.3343
0.0382	16.0	2416	0.3369
0.022	17.0	2567	0.3442
0.017	18.0	2718	0.3532
0.0286	19.0	2869	0.3594
0.0178	20.0	3020	0.3793
0.017	21.0	3171	0.3976
0.0127	22.0	3322	0.4029
0.0136	23.0	3473	0.4038
0.0124	24.0	3624	0.4045
0.0089	25.0	3775	0.3999
0.0108	26.0	3926	0.4103
0.0141	27.0	4077	0.4106
0.0126	28.0	4228	0.4197
0.0071	29.0	4379	0.4409
0.0046	30.0	4530	0.4240
0.0075	31.0	4681	0.4384
0.0061	32.0	4832	0.4313
0.0093	33.0	4983	0.4534
0.0066	34.0	5134	0.4499
0.0028	35.0	5285	0.4574
0.0034	36.0	5436	0.4615
0.0031	37.0	5587	0.4636
0.0047	38.0	5738	0.4728
0.0021	39.0	5889	0.4639
0.0027	40.0	6040	0.4658
0.0035	41.0	6191	0.4755
0.0032	42.0	6342	0.4589
0.0027	43.0	6493	0.4628
0.0018	44.0	6644	0.4762
0.0017	45.0	6795	0.4697
0.0012	46.0	6946	0.4762
0.0022	47.0	7097	0.4830
0.0013	48.0	7248	0.4765
0.0013	49.0	7399	0.4759
0.0014	50.0	7550	0.4773

Framework versions

• Transformers 4.41.1
• Pytorch 1.13.1+cu117
• Datasets 2.19.1
• Tokenizers 0.19.1