signon-project/mbart-large-cc25-ft-amr30-es

es+no_processing

This model is a fine-tuned version of facebook/mbart-large-cc25 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5874
• Smatch Precision: 74.08
• Smatch Recall: 76.84
• Smatch Fscore: 75.44
• Smatch Unparsable: 0
• Percent Not Recoverable: 0.2323

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: 5e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Smatch Precision	Smatch Recall	Smatch Fscore	Smatch Unparsable	Percent Not Recoverable
0.3908	1.0	3477	1.4300	19.74	68.95	30.7	0	0.0
0.256	2.0	6954	0.8998	27.75	70.61	39.85	1	0.0581
0.0704	3.0	10431	0.8727	30.09	72.2	42.47	0	0.1161
0.0586	4.0	13908	0.7774	37.1	74.93	49.62	0	0.1161
0.1059	5.0	17385	0.6322	42.52	74.54	54.15	1	0.1161
0.0424	6.0	20862	0.6090	47.13	76.21	58.25	0	0.0
0.0139	7.0	24339	0.5768	48.3	77.31	59.46	0	0.0581
0.08	8.0	27817	0.5608	55.74	77.16	64.72	0	0.1161
0.0224	9.0	31294	0.5937	54.91	77.02	64.11	0	0.0581
0.0757	10.0	34771	0.5588	59.53	77.47	67.32	0	0.0581
0.0613	11.0	38248	0.5894	60.83	77.82	68.28	0	0.0581
0.1045	12.0	41725	0.5847	61.23	77.17	68.28	0	0.1742
0.012	13.0	45202	0.5588	65.61	77.47	71.05	0	0.0
0.0591	14.0	48679	0.5609	66.51	77.86	71.74	0	0.0581
0.0252	15.0	52156	0.5653	67.48	77.75	72.25	0	0.0
0.0129	16.0	55634	0.5602	68.92	77.57	72.99	0	0.0
0.0006	17.0	59111	0.5876	68.57	77.81	72.9	0	0.1742
0.0182	18.0	62588	0.5951	68.97	77.96	73.19	0	0.1161
0.018	19.0	66065	0.5865	70.63	77.68	73.98	0	0.0581
0.0097	20.0	69542	0.6073	71.68	77.38	74.42	0	0.1161
0.0021	21.0	73019	0.5984	72.25	77.92	74.98	0	0.0581
0.0371	22.0	76496	0.5907	72.92	77.59	75.18	0	0.1742
0.0382	23.0	79973	0.5928	73.06	77.49	75.21	0	0.1742
0.0148	24.0	83451	0.5903	73.98	77.15	75.53	0	0.0581
0.1326	25.0	86925	0.5874	74.08	76.84	75.44	0	0.2323

Framework versions

• Transformers 4.34.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.14.2
• Tokenizers 0.13.3