49a5119ad6fe7f04ab1d69d7439106e9

This model is a fine-tuned version of facebook/opt-350m on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 1.2669
• Data Size: 1.0
• Epoch Runtime: 29.5056
• Accuracy: 0.7528
• F1 Macro: 0.7883
• Rouge1: 0.7528
• Rouge2: 0.0
• Rougel: 0.7528
• Rougelsum: 0.7528

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	2.8696	0	2.9179	0.2152	0.0998	0.2145	0.0	0.2152	0.2152
No log	1	178	4.5153	0.0078	3.8937	0.3587	0.1848	0.3587	0.0	0.3587	0.3580
No log	2	356	2.4650	0.0156	3.6617	0.4680	0.2835	0.4680	0.0	0.4688	0.4680
No log	3	534	0.8646	0.0312	4.4548	0.7045	0.5438	0.7053	0.0	0.7060	0.7045
No log	4	712	0.8895	0.0625	5.9351	0.7081	0.5501	0.7088	0.0	0.7088	0.7081
No log	5	890	0.8202	0.125	7.5084	0.6804	0.6824	0.6811	0.0	0.6811	0.6804
0.0628	6	1068	0.6393	0.25	10.7617	0.7528	0.7859	0.7543	0.0	0.7536	0.7528
0.5531	7	1246	0.6062	0.5	17.0990	0.7571	0.7865	0.7571	0.0	0.7571	0.7571
0.4495	8.0	1424	0.6959	1.0	30.0095	0.7457	0.7578	0.7464	0.0	0.7457	0.7457
0.2691	9.0	1602	0.8873	1.0	29.5465	0.7486	0.7400	0.7493	0.0	0.7486	0.7479
0.1516	10.0	1780	1.0456	1.0	30.4346	0.7393	0.7858	0.7393	0.0	0.7401	0.7386
0.1195	11.0	1958	1.2669	1.0	29.5056	0.7528	0.7883	0.7528	0.0	0.7528	0.7528

Framework versions

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