qa_kor_math_2

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

• Loss: 0.1234

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: 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.56	100	3.2887
No log	1.13	200	0.8359
No log	1.69	300	0.4944
No log	2.26	400	0.3843
2.4704	2.82	500	0.3349
2.4704	3.39	600	0.3005
2.4704	3.95	700	0.2768
2.4704	4.52	800	0.2641
2.4704	5.08	900	0.2479
0.3213	5.65	1000	0.2335
0.3213	6.21	1100	0.2208
0.3213	6.78	1200	0.2117
0.3213	7.34	1300	0.2041
0.3213	7.91	1400	0.1964
0.2503	8.47	1500	0.1876
0.2503	9.04	1600	0.1790
0.2503	9.6	1700	0.1745
0.2503	10.17	1800	0.1673
0.2503	10.73	1900	0.1623
0.2141	11.3	2000	0.1579
0.2141	11.86	2100	0.1527
0.2141	12.43	2200	0.1494
0.2141	12.99	2300	0.1438
0.2141	13.56	2400	0.1427
0.1873	14.12	2500	0.1386
0.1873	14.69	2600	0.1347
0.1873	15.25	2700	0.1334
0.1873	15.82	2800	0.1321
0.1873	16.38	2900	0.1295
0.1718	16.95	3000	0.1276
0.1718	17.51	3100	0.1263
0.1718	18.08	3200	0.1255
0.1718	18.64	3300	0.1244
0.1718	19.21	3400	0.1240
0.1628	19.77	3500	0.1234

Framework versions

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