marianmt-finetuned-netspeak-tgl-to-eng

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:

• Train Loss: 0.7277
• Validation Loss: 2.0459
• Train Bleu: 33.5501
• Train Gen Len: 8.7228
• Epoch: 93

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:

• optimizer: {'name': 'AdamWeightDecay', 'learning_rate': 2e-06, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01}
• training_precision: float32

Training results

Train Loss	Validation Loss	Train Bleu	Train Gen Len	Epoch
5.4267	4.5907	3.3310	11.7129	0
4.6862	4.1720	3.5594	10.4752	1
4.4077	3.9852	4.0100	9.2079	2
4.2296	3.8554	3.3190	9.3663	3
4.0964	3.7598	4.8776	9.4554	4
3.9799	3.6710	4.9744	9.6931	5
3.8799	3.5953	5.9838	9.4752	6
3.7661	3.5248	6.4073	9.3366	7
3.6807	3.4588	6.2692	9.1485	8
3.5932	3.3964	6.0781	9.0990	9
3.5110	3.3384	6.6363	9.0891	10
3.4294	3.2892	7.0472	9.2079	11
3.3566	3.2363	7.2707	9.1782	12
3.2796	3.1878	7.9426	9.1683	13
3.2026	3.1376	7.9254	9.2772	14
3.1472	3.0926	8.2076	9.1188	15
3.0634	3.0496	8.5193	9.2475	16
3.0124	3.0082	8.9990	9.1485	17
2.9554	2.9696	11.2816	9.1485	18
2.8885	2.9352	12.0866	9.0396	19
2.8403	2.8974	12.8611	9.1485	20
2.7636	2.8661	13.0981	9.1485	21
2.7229	2.8269	12.9295	8.9010	22
2.6714	2.7951	14.0159	8.8713	23
2.6179	2.7644	13.7369	8.7624	24
2.5520	2.7348	14.0979	8.8119	25
2.5199	2.7059	14.5253	8.7426	26
2.4652	2.6832	13.8452	8.7030	27
2.4081	2.6537	15.6475	8.9505	28
2.3708	2.6302	16.1325	8.8713	29
2.3195	2.6124	16.0044	8.7426	30
2.2938	2.5892	16.8560	8.8020	31
2.2202	2.5700	16.8995	8.8911	32
2.1808	2.5456	17.5342	8.8416	33
2.1373	2.5262	18.4092	8.6337	34
2.1096	2.5082	18.1906	8.6436	35
2.0610	2.4896	18.3189	8.7525	36
2.0275	2.4725	18.4318	8.6436	37
1.9913	2.4534	18.1136	8.6832	38
1.9544	2.4403	19.2999	8.6040	39
1.9144	2.4220	19.1325	8.6535	40
1.8781	2.4075	19.4122	8.6337	41
1.8610	2.3928	21.0270	8.6832	42
1.8176	2.3779	20.9122	8.7921	43
1.7839	2.3618	20.3906	8.7624	44
1.7553	2.3466	20.9078	8.7327	45
1.7045	2.3368	20.7228	8.7030	46
1.6974	2.3221	20.7889	8.7426	47
1.6561	2.3109	20.8293	8.7129	48
1.6264	2.2991	20.3201	8.5644	49
1.5976	2.2906	22.7905	8.6139	50
1.5725	2.2820	23.9301	8.7228	51
1.5528	2.2702	23.5437	8.6733	52
1.5158	2.2612	22.9832	8.6040	53
1.4883	2.2509	24.6290	8.6733	54
1.4497	2.2434	25.6293	8.6139	55
1.4357	2.2336	25.4158	8.6634	56
1.4105	2.2290	25.2337	8.5644	57
1.3803	2.2194	26.2588	8.5941	58
1.3606	2.2118	25.8251	8.6139	59
1.3389	2.2073	26.2269	8.5842	60
1.3064	2.1966	26.2973	8.6040	61
1.2747	2.1893	27.3831	8.5743	62
1.2586	2.1811	28.4823	8.6733	63
1.2445	2.1740	27.5688	8.6139	64
1.2201	2.1576	29.3111	8.5347	65
1.1924	2.1487	28.3428	8.6040	66
1.1657	2.1464	28.8596	8.5941	67
1.1435	2.1469	28.7870	8.5743	68
1.1274	2.1382	29.5455	8.6436	69
1.1080	2.1297	29.4602	8.6139	70
1.0907	2.1257	28.2800	8.7525	71
1.0881	2.1207	29.2731	8.6337	72
1.0534	2.1179	29.9292	8.7624	73
1.0389	2.1096	29.9660	8.5347	74
1.0186	2.1052	29.7106	8.5446	75
0.9953	2.0959	30.0563	8.5050	76
0.9727	2.0977	30.0527	8.5446	77
0.9543	2.0878	29.8762	8.5446	78
0.9372	2.0871	30.4451	8.4950	79
0.9234	2.0804	30.7829	8.5347	80
0.9045	2.0774	31.2911	8.6337	81
0.8920	2.0727	31.4189	8.4752	82
0.8729	2.0761	30.5640	8.7624	83
0.8466	2.0735	31.4347	8.7525	84
0.8430	2.0677	31.1463	8.6139	85
0.8340	2.0669	31.5623	8.7228	86
0.8152	2.0587	31.9364	8.6535	87
0.7916	2.0548	31.6855	8.6238	88
0.7829	2.0562	33.4523	8.7426	89
0.7678	2.0559	32.0304	8.7129	90
0.7509	2.0540	32.7711	8.7525	91
0.7406	2.0498	33.6200	8.7030	92
0.7277	2.0459	33.5501	8.7228	93

Framework versions

• Transformers 4.25.1
• TensorFlow 2.9.2
• Datasets 2.8.0
• Tokenizers 0.13.2