41b540e32dd13c7f922cdf0749d793cb

This model is a fine-tuned version of openai-community/gpt2-medium on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 1.5462
• Data Size: 1.0
• Epoch Runtime: 28.2172
• Accuracy: 0.7670
• F1 Macro: 0.8090
• Rouge1: 0.7670
• Rouge2: 0.0
• Rougel: 0.7678
• Rougelsum: 0.7670

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	8.8067	0	2.6919	0.2457	0.0797	0.2457	0.0	0.2457	0.2457
No log	1	178	3.4989	0.0078	4.2544	0.2180	0.1109	0.2180	0.0	0.2180	0.2173
No log	2	356	2.1025	0.0156	3.5522	0.2486	0.2133	0.2489	0.0	0.2486	0.2486
No log	3	534	1.2631	0.0312	4.6160	0.6357	0.4391	0.6364	0.0	0.6371	0.6349
No log	4	712	0.8732	0.0625	5.7011	0.7195	0.5607	0.7202	0.0	0.7202	0.7195
No log	5	890	0.7316	0.125	7.1848	0.7251	0.5604	0.7259	0.0	0.7259	0.7259
0.0852	6	1068	0.7241	0.25	10.7063	0.7102	0.5636	0.7102	0.0	0.7109	0.7102
0.5644	7	1246	0.6063	0.5	16.9841	0.7834	0.7771	0.7841	0.0	0.7841	0.7834
0.4466	8.0	1424	0.5941	1.0	28.8419	0.7841	0.8023	0.7848	0.0	0.7848	0.7848
0.2949	9.0	1602	0.7343	1.0	28.8525	0.7734	0.8095	0.7738	0.0	0.7741	0.7738
0.1573	10.0	1780	0.9434	1.0	28.7013	0.7926	0.8219	0.7926	0.0	0.7933	0.7933
0.1107	11.0	1958	1.2888	1.0	28.1867	0.7777	0.8142	0.7770	0.0	0.7777	0.7777
0.0603	12.0	2136	1.5462	1.0	28.2172	0.7670	0.8090	0.7670	0.0	0.7678	0.7670

Framework versions

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