0071116f8ad08e939efe84696374ecc2

This model is a fine-tuned version of distilbert/distilbert-base-uncased on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 1.3853
• Data Size: 1.0
• Epoch Runtime: 6.1902
• Accuracy: 0.7266
• F1 Macro: 0.7506
• Rouge1: 0.7266
• Rouge2: 0.0
• Rougel: 0.7266
• Rougelsum: 0.7259

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	1.6161	0	0.9691	0.1044	0.0678	0.1044	0.0	0.1044	0.1044
No log	1	178	1.5353	0.0078	1.4556	0.2493	0.0910	0.2493	0.0	0.25	0.2493
No log	2	356	1.4282	0.0156	1.4791	0.2791	0.1289	0.2791	0.0	0.2791	0.2791
No log	3	534	1.1495	0.0312	1.6470	0.6527	0.4477	0.6534	0.0	0.6534	0.6527
No log	4	712	0.8429	0.0625	1.9758	0.7074	0.5401	0.7081	0.0	0.7074	0.7081
No log	5	890	0.7985	0.125	1.9841	0.6960	0.5181	0.6967	0.0	0.6974	0.6960
0.0591	6	1068	0.7281	0.25	2.7029	0.7209	0.5589	0.7216	0.0	0.7216	0.7209
0.5657	7	1246	0.6417	0.5	3.6599	0.7557	0.7213	0.7564	0.0	0.7557	0.7550
0.4803	8.0	1424	0.5995	1.0	6.2900	0.7543	0.7515	0.7543	0.0	0.7550	0.7543
0.2774	9.0	1602	0.6802	1.0	6.2135	0.7720	0.7858	0.7727	0.0	0.7720	0.7720
0.1577	10.0	1780	0.8380	1.0	6.0862	0.7599	0.7671	0.7603	0.0	0.7599	0.7599
0.0915	11.0	1958	1.1013	1.0	6.1799	0.7585	0.7754	0.7585	0.0	0.7592	0.7578
0.0704	12.0	2136	1.3853	1.0	6.1902	0.7266	0.7506	0.7266	0.0	0.7266	0.7259

Framework versions

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