mt5-small-mt5-finetuned-final

This model is a fine-tuned version of google/mt5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1778
• Rouge1: 0.2833
• Rouge2: 0.1521
• Rougel: 0.2758
• Rougelsum: 0.2768

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: 0.0056
• train_batch_size: 12
• eval_batch_size: 12
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
5.9607	1.0	100	4.8449	0.1763	0.0684	0.1763	0.1761
4.9088	2.0	200	3.9878	0.3076	0.1348	0.2803	0.2815
2.9924	3.0	300	2.2397	0.2790	0.1378	0.2575	0.2592
2.2734	4.0	400	1.9866	0.2987	0.1629	0.2868	0.2872
1.9431	5.0	500	1.7408	0.2251	0.1380	0.2231	0.2237
2.317	6.0	600	1.9235	0.2421	0.0922	0.2276	0.2282
1.8526	7.0	700	1.6342	0.3120	0.1636	0.2943	0.2944
1.7029	8.0	800	1.6244	0.2469	0.1361	0.2421	0.2427
1.6725	9.0	900	1.5803	0.2637	0.1362	0.2551	0.2560
1.5852	10.0	1000	1.5617	0.2963	0.1634	0.2907	0.2917
1.4625	11.0	1100	1.4049	0.2750	0.1383	0.2570	0.2576
1.3895	12.0	1200	1.4234	0.2969	0.1646	0.2917	0.2927
1.3584	13.0	1300	1.3807	0.3370	0.1601	0.3088	0.3099
1.2759	14.0	1400	1.3524	0.2890	0.1307	0.2654	0.2663
1.222	15.0	1500	1.3110	0.2718	0.1339	0.2566	0.2597
1.1515	16.0	1600	1.2297	0.3314	0.1626	0.3033	0.3038
1.0888	17.0	1700	1.1897	0.3028	0.1358	0.2769	0.2792
1.039	18.0	1800	1.1970	0.2833	0.1521	0.2758	0.2768
0.9907	19.0	1900	1.1790	0.2833	0.1521	0.2758	0.2768
0.9563	20.0	2000	1.1778	0.2833	0.1521	0.2758	0.2768

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0