bananaspectre/marian-finetuned-tgl-eng-netspeak-trial9

Source: 🤗🏴‍☠️ Rescued by: isilicontranslationapache-2.0295 MBother✓ 6 checksumsupdated 1,406d ago

marian-finetuned-tgl-eng-netspeak-trial9

This model is a fine-tuned version of Helsinki-NLP/opus-mt-tl-en on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 2.3568
Bleu: 29.1370

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: 2e-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
num_epochs: 200
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu
5.6316	1.0	57	4.1183	4.6005
4.8654	2.0	114	3.8581	5.9435
4.5642	3.0	171	3.6865	6.6312
4.3364	4.0	228	3.5646	7.3157
4.1474	5.0	285	3.4686	8.6700
4.0044	6.0	342	3.3852	8.6733
3.8862	7.0	399	3.3257	8.3894
3.7676	8.0	456	3.2528	9.2599
3.6633	9.0	513	3.2005	9.7922
3.5594	10.0	570	3.1615	10.9836
3.4683	11.0	627	3.1055	11.0111
3.3897	12.0	684	3.0527	11.0658
3.3165	13.0	741	3.0106	11.3570
3.2338	14.0	798	2.9819	12.4296
3.1626	15.0	855	2.9395	13.0279
3.1127	16.0	912	2.9088	13.2959
3.0224	17.0	969	2.8760	13.9185
2.9523	18.0	1026	2.8420	14.7849
2.9036	19.0	1083	2.8059	15.3255
2.8449	20.0	1140	2.7830	15.8899
2.7851	21.0	1197	2.7654	15.3016
2.7182	22.0	1254	2.7422	15.9169
2.683	23.0	1311	2.7171	15.4695
2.6016	24.0	1368	2.6860	17.2504
2.5688	25.0	1425	2.6800	17.4693
2.511	26.0	1482	2.6523	17.8363
2.4627	27.0	1539	2.6247	18.6818
2.4259	28.0	1596	2.6038	19.2026
2.3814	29.0	1653	2.5946	18.9046
2.3368	30.0	1710	2.5720	19.6498
2.2639	31.0	1767	2.5564	18.7972
2.2366	32.0	1824	2.5432	20.1555
2.1884	33.0	1881	2.5369	19.9048
2.143	34.0	1938	2.5215	19.3706
2.122	35.0	1995	2.5102	20.2954
2.0819	36.0	2052	2.4966	20.5785
2.0333	37.0	2109	2.4939	20.8078
1.9972	38.0	2166	2.4852	21.6624
1.9596	39.0	2223	2.4724	21.3380
1.9386	40.0	2280	2.4550	21.7399
1.8881	41.0	2337	2.4539	21.8201
1.8488	42.0	2394	2.4494	22.8561
1.8344	43.0	2451	2.4436	22.0001
1.8005	44.0	2508	2.4353	21.5060
1.7703	45.0	2565	2.4314	22.6523
1.7321	46.0	2622	2.4258	22.9500
1.6897	47.0	2679	2.4202	23.0767
1.6822	48.0	2736	2.4115	23.3565
1.6392	49.0	2793	2.4056	24.4669
1.621	50.0	2850	2.4071	25.7900
1.6075	51.0	2907	2.3930	25.8570
1.5558	52.0	2964	2.3835	26.0207
1.5335	53.0	3021	2.3848	24.5089
1.5091	54.0	3078	2.3870	26.7579
1.4904	55.0	3135	2.3791	26.2250
1.4645	56.0	3192	2.3760	26.1819
1.4628	57.0	3249	2.3811	25.9747
1.4297	58.0	3306	2.3659	26.4407
1.4011	59.0	3363	2.3650	27.1145
1.3649	60.0	3420	2.3597	27.6616
1.3419	61.0	3477	2.3601	28.6248
1.3278	62.0	3534	2.3670	27.2075
1.3106	63.0	3591	2.3588	27.3917
1.2855	64.0	3648	2.3508	27.8277
1.2732	65.0	3705	2.3622	28.3032
1.259	66.0	3762	2.3603	28.0315
1.2397	67.0	3819	2.3551	27.9452
1.2285	68.0	3876	2.3597	28.5887
1.1898	69.0	3933	2.3599	28.5675
1.181	70.0	3990	2.3642	29.7412
1.1748	71.0	4047	2.3577	29.2003
1.146	72.0	4104	2.3609	28.3760
1.1274	73.0	4161	2.3519	29.2015
1.1138	74.0	4218	2.3568	29.1370

Framework versions

Transformers 4.25.1
Pytorch 1.13.0+cu116
Datasets 2.8.0
Tokenizers 0.13.2