dyu-fr-helsinki_v1

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

• Loss: 4.9661
• Bleu score: 0.0816

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: 16
• eval_batch_size: 20
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu score
4.0349	1.0	505	3.5486	0.0063
3.4137	2.0	1010	3.3835	0.0148
3.035	3.0	1515	3.2874	0.0250
2.7464	4.0	2020	3.2157	0.0309
2.4679	5.0	2525	3.1806	0.0338
2.2416	6.0	3030	3.1661	0.0458
2.0367	7.0	3535	3.1695	0.0534
1.862	8.0	4040	3.1763	0.0517
1.6801	9.0	4545	3.1986	0.0707
1.5227	10.0	5050	3.2324	0.0631
1.3919	11.0	5555	3.2623	0.0663
1.2575	12.0	6060	3.3066	0.0710
1.1428	13.0	6565	3.3293	0.0773
1.0335	14.0	7070	3.3601	0.0754
0.9396	15.0	7575	3.4277	0.0805
0.837	16.0	8080	3.4645	0.0829
0.7519	17.0	8585	3.5172	0.0875
0.6839	18.0	9090	3.5741	0.0828
0.606	19.0	9595	3.5909	0.0776
0.549	20.0	10100	3.6561	0.0783
0.4882	21.0	10605	3.6974	0.0814
0.4418	22.0	11110	3.7301	0.0816
0.394	23.0	11615	3.7658	0.0832
0.3529	24.0	12120	3.8177	0.0730
0.3143	25.0	12625	3.8601	0.0763
0.2806	26.0	13130	3.9042	0.0796
0.2573	27.0	13635	3.9459	0.0836
0.227	28.0	14140	3.9954	0.0762
0.2071	29.0	14645	4.0197	0.0814
0.1847	30.0	15150	4.0565	0.0743
0.1713	31.0	15655	4.0777	0.0756
0.1522	32.0	16160	4.1300	0.0763
0.1393	33.0	16665	4.1759	0.0829
0.1281	34.0	17170	4.1883	0.0773
0.1176	35.0	17675	4.2146	0.0752
0.1086	36.0	18180	4.2426	0.0807
0.1004	37.0	18685	4.2829	0.0812
0.0918	38.0	19190	4.3058	0.0818
0.0852	39.0	19695	4.3172	0.0788
0.0814	40.0	20200	4.3668	0.0738
0.074	41.0	20705	4.3948	0.0809
0.0697	42.0	21210	4.4054	0.0773
0.0639	43.0	21715	4.4292	0.0813
0.0612	44.0	22220	4.4613	0.0821
0.0588	45.0	22725	4.4713	0.0779
0.0539	46.0	23230	4.4898	0.0781
0.051	47.0	23735	4.5162	0.0799
0.0469	48.0	24240	4.5198	0.0791
0.0463	49.0	24745	4.5537	0.0777
0.0435	50.0	25250	4.5688	0.0771
0.0421	51.0	25755	4.5640	0.0776
0.038	52.0	26260	4.6035	0.0850
0.0375	53.0	26765	4.6087	0.0762
0.0342	54.0	27270	4.6410	0.0791
0.0333	55.0	27775	4.6573	0.0795
0.0314	56.0	28280	4.6953	0.0796
0.0302	57.0	28785	4.6893	0.0768
0.0287	58.0	29290	4.7088	0.0798
0.0282	59.0	29795	4.7017	0.0792
0.0254	60.0	30300	4.7246	0.0817
0.0254	61.0	30805	4.7453	0.0843
0.0242	62.0	31310	4.7498	0.0835
0.0237	63.0	31815	4.7742	0.0848
0.0232	64.0	32320	4.8010	0.0837
0.0206	65.0	32825	4.8000	0.0779
0.021	66.0	33330	4.8096	0.0816
0.0199	67.0	33835	4.8005	0.0807
0.0191	68.0	34340	4.8166	0.0797
0.019	69.0	34845	4.8154	0.0806
0.0186	70.0	35350	4.8536	0.0812
0.0175	71.0	35855	4.8440	0.0838
0.017	72.0	36360	4.8637	0.0833
0.016	73.0	36865	4.8559	0.0795
0.0159	74.0	37370	4.8711	0.0819
0.0162	75.0	37875	4.8803	0.0808
0.0152	76.0	38380	4.8949	0.0853
0.0137	77.0	38885	4.9082	0.0860
0.0141	78.0	39390	4.9174	0.0885
0.0135	79.0	39895	4.9045	0.0830
0.0133	80.0	40400	4.9047	0.0833
0.0129	81.0	40905	4.9136	0.0835
0.0128	82.0	41410	4.9245	0.0804
0.0118	83.0	41915	4.9271	0.0855
0.0117	84.0	42420	4.9248	0.0832
0.012	85.0	42925	4.9390	0.0852
0.0111	86.0	43430	4.9364	0.0828
0.0116	87.0	43935	4.9447	0.0827
0.0109	88.0	44440	4.9489	0.0814
0.0112	89.0	44945	4.9424	0.0818
0.0105	90.0	45450	4.9593	0.0832
0.0101	91.0	45955	4.9714	0.0853
0.0105	92.0	46460	4.9578	0.0851
0.01	93.0	46965	4.9532	0.0840
0.0095	94.0	47470	4.9586	0.0830
0.0094	95.0	47975	4.9648	0.0816
0.0096	96.0	48480	4.9753	0.0839
0.0095	97.0	48985	4.9681	0.0824
0.0092	98.0	49490	4.9688	0.0837
0.0089	99.0	49995	4.9656	0.0834
0.009	100.0	50500	4.9661	0.0816

Framework versions

• Transformers 4.44.0
• Pytorch 2.1.0+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1