542dfd15cb6065504b3669a10a059361

This model is a fine-tuned version of studio-ousia/luke-large on the contemmcm/trec dataset. It achieves the following results on the evaluation set:

• Loss: 1.7045
• Data Size: 0.5
• Epoch Runtime: 21.9134
• Accuracy: 0.1792
• F1 Macro: 0.0506

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
No log	0	0	1.8395	0	1.5456	0.2771	0.0723
No log	1	170	1.8069	0.0078	2.3311	0.2771	0.0723
No log	2	340	1.7485	0.0156	2.6148	0.2458	0.0991
No log	3	510	1.6337	0.0312	3.8700	0.1354	0.0423
No log	4	680	1.6808	0.0625	5.5459	0.1917	0.0686
0.1013	5	850	1.7271	0.125	8.3869	0.1792	0.0506
0.1013	6	1020	1.7025	0.25	12.8953	0.1792	0.0506
1.6806	7	1190	1.7045	0.5	21.9134	0.1792	0.0506

Framework versions

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