qa_kor_market

This model is a fine-tuned version of hyunwoongko/kobart on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7618

Model description

슈퍼 마켓에서 있을 법한한 문의 내용을 입력하면, 문의 의도, 문의 항목, 답변을 리턴해주는 모델입니다.

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: 1e-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
• lr_scheduler_warmup_steps: 400
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.03	100	3.4839
No log	0.05	200	1.4909
No log	0.08	300	1.2606
No log	0.1	400	1.1675
3.0259	0.13	500	1.1008
3.0259	0.15	600	1.0580
3.0259	0.18	700	1.0222
3.0259	0.2	800	0.9938
3.0259	0.23	900	0.9707
1.0853	0.25	1000	0.9571
1.0853	0.28	1100	0.9370
1.0853	0.3	1200	0.9293
1.0853	0.33	1300	0.9146
1.0853	0.35	1400	0.9065
0.992	0.38	1500	0.8997
0.992	0.41	1600	0.8930
0.992	0.43	1700	0.8834
0.992	0.46	1800	0.8788
0.992	0.48	1900	0.8714
0.9418	0.51	2000	0.8706
0.9418	0.53	2100	0.8676
0.9418	0.56	2200	0.8619
0.9418	0.58	2300	0.8548
0.9418	0.61	2400	0.8514
0.9222	0.63	2500	0.8511
0.9222	0.66	2600	0.8483
0.9222	0.68	2700	0.8425
0.9222	0.71	2800	0.8396
0.9222	0.74	2900	0.8384
0.8981	0.76	3000	0.8360
0.8981	0.79	3100	0.8295
0.8981	0.81	3200	0.8290
0.8981	0.84	3300	0.8273
0.8981	0.86	3400	0.8221
0.874	0.89	3500	0.8228
0.874	0.91	3600	0.8213
0.874	0.94	3700	0.8183
0.874	0.96	3800	0.8163
0.874	0.99	3900	0.8178
0.8575	1.01	4000	0.8143
0.8575	1.04	4100	0.8118
0.8575	1.06	4200	0.8094
0.8575	1.09	4300	0.8092
0.8575	1.12	4400	0.8085
0.8374	1.14	4500	0.8048
0.8374	1.17	4600	0.8041
0.8374	1.19	4700	0.8018
0.8374	1.22	4800	0.8007
0.8374	1.24	4900	0.7988
0.8282	1.27	5000	0.7980
0.8282	1.29	5100	0.7968
0.8282	1.32	5200	0.7974
0.8282	1.34	5300	0.7949
0.8282	1.37	5400	0.7919
0.8149	1.39	5500	0.7931
0.8149	1.42	5600	0.7900
0.8149	1.44	5700	0.7887
0.8149	1.47	5800	0.7875
0.8149	1.5	5900	0.7883
0.8098	1.52	6000	0.7886
0.8098	1.55	6100	0.7860
0.8098	1.57	6200	0.7873
0.8098	1.6	6300	0.7822
0.8098	1.62	6400	0.7841
0.8306	1.65	6500	0.7828
0.8306	1.67	6600	0.7817
0.8306	1.7	6700	0.7812
0.8306	1.72	6800	0.7814
0.8306	1.75	6900	0.7799
0.7974	1.77	7000	0.7774
0.7974	1.8	7100	0.7795
0.7974	1.83	7200	0.7782
0.7974	1.85	7300	0.7786
0.7974	1.88	7400	0.7773
0.7945	1.9	7500	0.7749
0.7945	1.93	7600	0.7737
0.7945	1.95	7700	0.7743
0.7945	1.98	7800	0.7742
0.7945	2.0	7900	0.7732
0.8005	2.03	8000	0.7758
0.8005	2.05	8100	0.7726
0.8005	2.08	8200	0.7716
0.8005	2.1	8300	0.7742
0.8005	2.13	8400	0.7720
0.7788	2.15	8500	0.7706
0.7788	2.18	8600	0.7701
0.7788	2.21	8700	0.7702
0.7788	2.23	8800	0.7676
0.7788	2.26	8900	0.7699
0.7685	2.28	9000	0.7689
0.7685	2.31	9100	0.7677
0.7685	2.33	9200	0.7686
0.7685	2.36	9300	0.7671
0.7685	2.38	9400	0.7668
0.7814	2.41	9500	0.7670
0.7814	2.43	9600	0.7669
0.7814	2.46	9700	0.7661
0.7814	2.48	9800	0.7653
0.7814	2.51	9900	0.7663
0.7824	2.53	10000	0.7655
0.7824	2.56	10100	0.7654
0.7824	2.59	10200	0.7653
0.7824	2.61	10300	0.7652
0.7824	2.64	10400	0.7640
0.7798	2.66	10500	0.7647
0.7798	2.69	10600	0.7637
0.7798	2.71	10700	0.7636
0.7798	2.74	10800	0.7629
0.7798	2.76	10900	0.7629
0.7619	2.79	11000	0.7629
0.7619	2.81	11100	0.7624
0.7619	2.84	11200	0.7621
0.7619	2.86	11300	0.7621
0.7619	2.89	11400	0.7623
0.7723	2.92	11500	0.7621
0.7723	2.94	11600	0.7619
0.7723	2.97	11700	0.7619
0.7723	2.99	11800	0.7618

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2