opus-mt-tc-big-en-ar-finetuned-franco-to-arabic-1

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

• Loss: 2.1589
• Bleu: 27.737
• Gen Len: 20.9565

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

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	13	3.2371	9.9987	21.0
No log	2.0	26	3.1023	10.8123	19.7826
No log	3.0	39	2.9474	9.0815	20.4783
No log	4.0	52	2.8274	11.6434	21.3478
No log	5.0	65	2.6879	11.6494	21.4348
No log	6.0	78	2.6243	16.1661	21.2174
No log	7.0	91	2.5342	16.5707	22.1304
No log	8.0	104	2.5489	17.885	20.2609
No log	9.0	117	2.4488	17.6415	20.4348
No log	10.0	130	2.4177	18.5943	20.5652
No log	11.0	143	2.3239	20.0857	21.3043
No log	12.0	156	2.2723	22.9544	21.4348
No log	13.0	169	2.2456	25.4455	21.5217
No log	14.0	182	2.2628	24.8757	21.2609
No log	15.0	195	2.2229	26.0051	21.3478
No log	16.0	208	2.2090	27.6742	20.8261
No log	17.0	221	2.1762	26.7048	21.3913
No log	18.0	234	2.1786	27.7328	21.1304
No log	19.0	247	2.1631	27.7187	21.2174
No log	20.0	260	2.1589	27.737	20.9565

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Tokenizers 0.13.3