fcbecc6d81cf2bab82864810604f6b57

This model is a fine-tuned version of openai-community/gpt2-xl on the nyu-mll/glue [mnli] dataset. It achieves the following results on the evaluation set:

• Loss: 0.7080
• Data Size: 1.0
• Epoch Runtime: 3303.9896
• Accuracy: 0.8028
• F1 Macro: 0.8029
• Rouge1: 0.8027
• Rouge2: 0.0
• Rougel: 0.8028
• Rougelsum: 0.8027

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	1.1737	0	27.9329	0.3439	0.2808	0.3435	0.0	0.3438	0.3440
1.0046	1	12271	0.6957	0.0078	54.1598	0.7116	0.7119	0.7116	0.0	0.7118	0.7119
0.6586	2	24542	0.5895	0.0156	81.5933	0.7597	0.7597	0.7599	0.0	0.7597	0.7598
0.5868	3	36813	0.5331	0.0312	134.6484	0.7857	0.7836	0.7856	0.0	0.7858	0.7856
0.5504	4	49084	0.5104	0.0625	235.4530	0.7913	0.7917	0.7913	0.0	0.7913	0.7912
0.5119	5	61355	0.5149	0.125	439.6727	0.7926	0.7916	0.7924	0.0	0.7928	0.7928
0.5064	6	73626	0.5197	0.25	850.4059	0.7987	0.7983	0.7986	0.0	0.7987	0.7986
0.4213	7	85897	0.5078	0.5	1674.6778	0.8028	0.8019	0.8026	0.0	0.8028	0.8029
0.3938	8.0	98168	0.5165	1.0	3305.2754	0.8038	0.8044	0.8036	0.0	0.8039	0.8038
0.2647	9.0	110439	0.5350	1.0	3422.1590	0.8118	0.8104	0.8116	0.0	0.8119	0.8119
0.2338	10.0	122710	0.5871	1.0	3369.2638	0.8067	0.8054	0.8068	0.0	0.8065	0.8069
0.1867	11.0	134981	0.7080	1.0	3303.9896	0.8028	0.8029	0.8027	0.0	0.8028	0.8027

Framework versions

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