261e90d1a5fbff63aab7f80d4d74054f

This model is a fine-tuned version of studio-ousia/mluke-base on the nyu-mll/glue [qqp] dataset. It achieves the following results on the evaluation set:

• Loss: 0.2879
• Data Size: 1.0
• Epoch Runtime: 1039.3717
• Accuracy: 0.8901
• F1 Macro: 0.8845
• Rouge1: 0.8900
• Rouge2: 0.0
• Rougel: 0.8901
• Rougelsum: 0.8901

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	0.6963	0	32.4334	0.4860	0.4853	0.4861	0.0	0.4860	0.4862
0.5565	1	11370	0.5268	0.0078	40.5925	0.7624	0.7204	0.7623	0.0	0.7624	0.7623
0.4197	2	22740	0.4244	0.0156	49.3222	0.8088	0.7988	0.8088	0.0	0.8087	0.8087
0.3983	3	34110	0.3950	0.0312	65.2525	0.8290	0.8185	0.8289	0.0	0.8291	0.8289
0.3634	4	45480	0.3648	0.0625	95.8633	0.8425	0.8292	0.8424	0.0	0.8425	0.8425
0.3397	5	56850	0.3159	0.125	158.8982	0.8580	0.8510	0.8580	0.0	0.8580	0.8580
0.3217	6	68220	0.3098	0.25	282.5403	0.8564	0.8511	0.8565	0.0	0.8565	0.8563
0.2621	7	79590	0.2727	0.5	528.8700	0.8802	0.8738	0.8802	0.0	0.8803	0.8802
0.291	8.0	90960	0.2620	1.0	1036.0359	0.8890	0.8799	0.8890	0.0	0.8891	0.8891
0.2484	9.0	102330	0.2836	1.0	1035.2419	0.8889	0.8829	0.8890	0.0	0.8889	0.8890
0.2184	10.0	113700	0.2849	1.0	1032.5327	0.8921	0.8844	0.8920	0.0	0.8922	0.8921
0.2234	11.0	125070	0.3024	1.0	1028.5292	0.8847	0.8782	0.8847	0.0	0.8846	0.8847
0.1823	12.0	136440	0.2879	1.0	1039.3717	0.8901	0.8845	0.8900	0.0	0.8901	0.8901

Framework versions

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