bfd7277844e1eb553316f2d93f8d3b5a

This model is a fine-tuned version of distilbert/distilgpt2 on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 0.8602
• Data Size: 1.0
• Epoch Runtime: 8.8766
• Accuracy: 0.7614
• F1 Macro: 0.7837
• Rouge1: 0.7621
• Rouge2: 0.0
• Rougel: 0.7621
• Rougelsum: 0.7621

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	5.6572	0	1.3577	0.2166	0.0714	0.2159	0.0	0.2166	0.2166
No log	1	178	2.6240	0.0078	1.6841	0.1911	0.1129	0.1911	0.0	0.1911	0.1903
No log	2	356	2.0803	0.0156	1.7959	0.4318	0.2501	0.4332	0.0	0.4325	0.4318
No log	3	534	1.3150	0.0312	2.0462	0.4489	0.3189	0.4496	0.0	0.4489	0.4482
No log	4	712	0.9142	0.0625	2.2772	0.6527	0.5066	0.6534	0.0	0.6531	0.6520
No log	5	890	0.8002	0.125	2.5327	0.7116	0.5491	0.7131	0.0	0.7124	0.7116
0.073	6	1068	0.6805	0.25	3.4066	0.7351	0.5885	0.7358	0.0	0.7358	0.7351
0.6183	7	1246	0.6658	0.5	5.1568	0.7571	0.5976	0.7578	0.0	0.7571	0.7564
0.5409	8.0	1424	0.5889	1.0	8.7451	0.7642	0.7515	0.7649	0.0	0.7642	0.7642
0.4289	9.0	1602	0.6296	1.0	8.6249	0.7649	0.7767	0.7649	0.0	0.7656	0.7649
0.3434	10.0	1780	0.6846	1.0	8.9443	0.7536	0.7516	0.7536	0.0	0.7543	0.7543
0.2615	11.0	1958	0.7621	1.0	8.7761	0.7670	0.7741	0.7670	0.0	0.7678	0.7670
0.1575	12.0	2136	0.8602	1.0	8.8766	0.7614	0.7837	0.7621	0.0	0.7621	0.7621

Framework versions

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