README.md

ALERT!!!

leadawon/jeju-ko-nmt-v6 is better than leadawon/jeju-ko-nmt-v8

6버전 성능이 더 좋습니다!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

tag

---

tags: - generated_from_trainer model-index: - name: jeju-ko-nmt-v8 results: []

jeju-ko-nmt-v8

This model is a fine-tuned version of leadawon/jeju-ko-nmt-v7 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2448

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-06
• train_batch_size: 24
• eval_batch_size: 24
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 96
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.2684	0.04	500	0.2568
0.2468	0.08	1000	0.2547
0.2167	0.12	1500	0.2540
0.1966	0.16	2000	0.2535
0.1846	0.2	2500	0.2533
0.1727	0.24	3000	0.2535
0.1746	0.28	3500	0.2522
0.1726	0.32	4000	0.2521
0.1722	0.36	4500	0.2519
0.1731	0.4	5000	0.2515
0.1701	0.44	5500	0.2518
0.168	0.48	6000	0.2515
0.1706	0.52	6500	0.2509
0.1659	0.56	7000	0.2514
0.1702	0.6	7500	0.2509
0.1667	0.64	8000	0.2510
0.1661	0.68	8500	0.2508
0.1647	0.72	9000	0.2510
0.1632	0.76	9500	0.2510
0.1655	0.8	10000	0.2506
0.1645	0.84	10500	0.2508
0.1617	0.88	11000	0.2508
0.1627	0.91	11500	0.2511
0.2764	0.95	12000	0.2478
0.2755	0.99	12500	0.2462
0.2275	1.03	13000	0.2464
0.2201	1.07	13500	0.2463
0.2207	1.11	14000	0.2463
0.2202	1.15	14500	0.2462
0.2194	1.19	15000	0.2460
0.2177	1.23	15500	0.2461
0.2187	1.27	16000	0.2460
0.2184	1.31	16500	0.2459
0.2182	1.35	17000	0.2457
0.219	1.39	17500	0.2458
0.2206	1.43	18000	0.2455
0.2211	1.47	18500	0.2455
0.2164	1.51	19000	0.2455
0.2202	1.55	19500	0.2454
0.2208	1.59	20000	0.2452
0.2208	1.63	20500	0.2450
0.2204	1.67	21000	0.2450
0.2193	1.71	21500	0.2450
0.221	1.75	22000	0.2451
0.2168	1.79	22500	0.2450
0.2169	1.83	23000	0.2449
0.218	1.87	23500	0.2449
0.2196	1.91	24000	0.2449
0.2218	1.95	24500	0.2448
0.2199	1.99	25000	0.2448

Framework versions

• Transformers 4.26.0
• Pytorch 1.13.1+cu116
• Tokenizers 0.13.2