0169b911bc1ab7b60683e4cb0de84cf6

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

• Loss: 1.1305
• Data Size: 1.0
• Epoch Runtime: 8.4361
• Accuracy: 0.7649
• F1 Macro: 0.7906
• Rouge1: 0.7649
• Rouge2: 0.0
• Rougel: 0.7656
• Rougelsum: 0.7656

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	1.6420	0	1.2008	0.0213	0.0113	0.0213	0.0	0.0206	0.0213
No log	1	178	1.5783	0.0078	2.6539	0.3253	0.1625	0.3246	0.0	0.3246	0.3253
No log	2	356	1.4686	0.0156	1.4772	0.3828	0.2148	0.3835	0.0	0.3835	0.3828
No log	3	534	1.0581	0.0312	1.7015	0.6044	0.4460	0.6044	0.0	0.6044	0.6051
No log	4	712	0.8228	0.0625	1.9030	0.7045	0.5533	0.7060	0.0	0.7053	0.7045
No log	5	890	0.7139	0.125	2.3489	0.7365	0.5834	0.7365	0.0	0.7372	0.7358
0.056	6	1068	0.7088	0.25	3.1549	0.7230	0.5777	0.7237	0.0	0.7237	0.7230
0.5765	7	1246	0.6390	0.5	4.8807	0.7585	0.7443	0.7592	0.0	0.7592	0.7592
0.4955	8.0	1424	0.5722	1.0	8.5434	0.7791	0.8038	0.7791	0.0	0.7798	0.7798
0.3464	9.0	1602	0.6713	1.0	8.5169	0.7635	0.8019	0.7635	0.0	0.7649	0.7642
0.243	10.0	1780	0.8033	1.0	8.3131	0.7578	0.7945	0.7585	0.0	0.7585	0.7585
0.174	11.0	1958	0.9666	1.0	8.5278	0.7763	0.8031	0.7770	0.0	0.7766	0.7770
0.1218	12.0	2136	1.1305	1.0	8.4361	0.7649	0.7906	0.7649	0.0	0.7656	0.7656

Framework versions

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