8b3d899cd30a76cfaa28ed4fc354077a

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

• Loss: 0.4000
• Data Size: 1.0
• Epoch Runtime: 290.7913
• Accuracy: 0.8904
• F1 Macro: 0.8903
• Rouge1: 0.8904
• Rouge2: 0.0
• Rougel: 0.8906
• Rougelsum: 0.8904

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	0.6923	0	5.1116	0.5132	0.5132	0.5136	0.0	0.5132	0.5132
No log	1	3273	0.5711	0.0078	8.8252	0.7507	0.7505	0.7506	0.0	0.7506	0.7504
0.01	2	6546	0.6778	0.0156	10.5576	0.6917	0.6629	0.6915	0.0	0.6914	0.6915
0.5304	3	9819	0.4251	0.0312	15.3382	0.8195	0.8191	0.8195	0.0	0.8193	0.8195
0.4628	4	13092	0.3867	0.0625	24.6712	0.8347	0.8346	0.8346	0.0	0.8349	0.8347
0.3748	5	16365	0.3535	0.125	42.2870	0.8546	0.8546	0.8544	0.0	0.8548	0.8544
0.3661	6	19638	0.3130	0.25	77.8438	0.8693	0.8693	0.8694	0.0	0.8693	0.8695
0.3253	7	22911	0.2789	0.5	149.8575	0.8847	0.8845	0.8846	0.0	0.8849	0.8846
0.3055	8.0	26184	0.2840	1.0	291.3508	0.8862	0.8862	0.8864	0.0	0.8860	0.8862
0.2233	9.0	29457	0.2958	1.0	294.5127	0.8901	0.8900	0.8901	0.0	0.8903	0.8899
0.1823	10.0	32730	0.2922	1.0	292.0687	0.8952	0.8951	0.8950	0.0	0.8952	0.8954
0.1455	11.0	36003	0.4000	1.0	290.7913	0.8904	0.8903	0.8904	0.0	0.8906	0.8904

Framework versions

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