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

en+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: 0.4481
• Smatch Precision: 80.57
• Smatch Recall: 83.81
• Smatch Fscore: 82.16
• Smatch Unparsable: 0
• Percent Not Recoverable: 0.3484

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
0.3471	1.0	3477	1.4889	22.35	73.05	34.23	0	0.1161
0.1741	2.0	6954	0.8681	30.1	71.92	42.44	0	0.1161
0.1296	3.0	10431	0.7081	38.6	78.68	51.8	0	0.0581
0.1308	4.0	13908	0.9546	37.49	78.23	50.69	0	0.0
0.2213	5.0	17385	0.5544	47.63	81.17	60.03	0	0.0
0.0317	6.0	20862	0.4884	49.3	80.9	61.27	0	0.0
0.1007	7.0	24339	0.4763	54.88	82.09	65.78	0	0.0
0.092	8.0	27817	0.4444	57.37	83.2	67.91	0	0.0
0.1051	9.0	31294	0.4192	64.37	83.81	72.82	0	0.0
0.0079	10.0	34771	0.4685	61.3	83.1	70.55	0	0.0
0.0211	11.0	38248	0.4389	63.36	84.57	72.44	0	0.1161
0.1122	12.0	41725	0.4146	69.39	83.56	75.82	0	0.0581
0.0183	13.0	45202	0.4003	73.9	83.71	78.5	0	0.0
0.0244	14.0	48679	0.4208	73.79	83.92	78.53	0	0.1161
0.0116	15.0	52156	0.4248	73.88	83.85	78.55	0	0.1161
0.0357	16.0	55634	0.4235	75.78	84.08	79.71	0	0.1161
0.0006	17.0	59111	0.4181	76.15	84.15	79.95	0	0.0581
0.0329	18.0	62588	0.4494	77.21	84.12	80.52	0	0.0
0.0003	19.0	66065	0.4389	78.02	84.13	80.96	0	0.0
0.04	20.0	69542	0.4439	78.78	84.23	81.41	0	0.0
0.0182	21.0	73019	0.4430	79.82	84.05	81.88	0	0.0581
0.0006	22.0	76496	0.4488	79.96	83.74	81.81	0	0.0581
0.0074	23.0	79973	0.4569	79.84	83.85	81.79	0	0.0581
0.0133	24.0	83451	0.4469	80.45	83.81	82.09	0	0.2904
0.0055	25.0	86925	0.4481	80.57	83.81	82.16	0	0.3484

Framework versions

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