output

This model is a fine-tuned version of facebook/m2m100_418M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2678
• Precision: 0.9166
• Recall: 0.9166
• F1 score: 0.9164
• Bleu: 28.7416
• Meteor: 0.3773
• Ter: 71.3529
• Chrf: 50.0450

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 800
• training_steps: 6000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1 score	Bleu	Meteor	Ter	Chrf
6.3055	0.7752	200	5.3503	0.8494	0.8425	0.8448	2.8881	0.1997	335.5156	20.8502
3.2698	1.5504	400	2.2718	0.9058	0.8907	0.8978	14.8848	0.3054	79.7919	39.0690
0.575	2.3256	600	0.4124	0.8988	0.9048	0.9014	13.4928	0.3267	90.0851	42.4825
0.3216	3.1008	800	0.3123	0.9062	0.9044	0.9050	17.0882	0.3180	80.1514	42.4666
0.2668	3.8760	1000	0.2778	0.9085	0.9021	0.9050	18.7200	0.3327	76.5374	43.3363
0.2141	4.6512	1200	0.2632	0.9112	0.9045	0.9075	19.1642	0.3293	75.0615	42.6765
0.161	5.4264	1400	0.2564	0.9130	0.9046	0.9085	21.6492	0.3430	72.2233	45.1452
0.1483	6.2016	1600	0.2493	0.9162	0.9088	0.9122	22.5441	0.3454	72.3746	45.8491
0.1314	6.9767	1800	0.2485	0.9151	0.9127	0.9136	24.5229	0.3622	73.7938	46.8310
0.1095	7.7519	2000	0.2467	0.9203	0.9099	0.9148	26.3288	0.3681	69.8392	47.8396
0.0883	8.5271	2200	0.2531	0.9181	0.9126	0.9151	26.4899	0.3657	70.3879	48.2937
0.0676	9.3023	2400	0.2560	0.9209	0.9120	0.9161	27.8690	0.3734	68.4201	49.1044
0.0693	10.0775	2600	0.2610	0.9162	0.9156	0.9156	27.6786	0.3756	71.6556	49.3039
0.0603	10.8527	2800	0.2624	0.9180	0.9156	0.9166	29.1587	0.3849	70.4257	50.4754
0.0468	11.6279	3000	0.2678	0.9166	0.9166	0.9164	28.7416	0.3773	71.3529	50.0450

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1