fb9677c7aa927b8923aa1ff350a7998d

This model is a fine-tuned version of FacebookAI/xlm-roberta-large on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 0.8891
• Data Size: 1.0
• Epoch Runtime: 32.7165
• Accuracy: 0.7670
• F1 Macro: 0.7776
• Rouge1: 0.7678
• Rouge2: 0.0
• Rougel: 0.7670
• Rougelsum: 0.7678

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	1.6469	0	2.6886	0.2401	0.0774	0.2401	0.0	0.2408	0.2393
No log	1	178	1.7121	0.0078	3.8238	0.4339	0.2328	0.4339	0.0	0.4339	0.4339
No log	2	356	1.4945	0.0156	3.7883	0.2486	0.0796	0.2486	0.0	0.2486	0.2479
No log	3	534	1.2110	0.0312	4.8726	0.5277	0.3716	0.5270	0.0	0.5270	0.5270
No log	4	712	0.8713	0.0625	6.4553	0.6847	0.5243	0.6854	0.0	0.6854	0.6847
No log	5	890	0.9573	0.125	8.3573	0.7209	0.5590	0.7216	0.0	0.7216	0.7209
0.0644	6	1068	0.8326	0.25	11.9767	0.7280	0.5685	0.7287	0.0	0.7294	0.7280
0.6871	7	1246	0.7736	0.5	19.0973	0.7266	0.5529	0.7273	0.0	0.7273	0.7266
0.5937	8.0	1424	0.6605	1.0	34.4357	0.75	0.7203	0.75	0.0	0.75	0.75
0.5084	9.0	1602	0.6172	1.0	32.7375	0.7784	0.8065	0.7791	0.0	0.7791	0.7791
0.4563	10.0	1780	0.6487	1.0	32.6570	0.7692	0.7920	0.7699	0.0	0.7692	0.7692
0.3482	11.0	1958	0.7184	1.0	32.5770	0.7337	0.7772	0.7337	0.0	0.7344	0.7337
0.2665	12.0	2136	0.8521	1.0	32.7439	0.7805	0.8050	0.7812	0.0	0.7812	0.7805
0.2552	13.0	2314	0.8891	1.0	32.7165	0.7670	0.7776	0.7678	0.0	0.7670	0.7678

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1