221cac15251d831995a3f3ac04e83452

This model is a fine-tuned version of facebook/opt-2.7b on the nyu-mll/glue [mrpc] dataset. It achieves the following results on the evaluation set:

• Loss: 2.0856
• Data Size: 1.0
• Epoch Runtime: 43.4524
• Accuracy: 0.6427
• F1 Macro: 0.5816
• Rouge1: 0.6427
• Rouge2: 0.0
• Rougel: 0.6436
• Rougelsum: 0.6427

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	0.8576	0	4.3545	0.4581	0.4555	0.4581	0.0	0.4581	0.4584
No log	1	114	2.9258	0.0078	5.8741	0.6657	0.4013	0.6663	0.0	0.6657	0.6657
No log	2	228	0.7741	0.0156	7.7419	0.5802	0.5363	0.5802	0.0	0.5790	0.5790
No log	3	342	1.6906	0.0312	14.0815	0.6798	0.4580	0.6804	0.0	0.6798	0.6795
0.0372	4	456	0.6361	0.0625	17.4017	0.6651	0.3994	0.6657	0.0	0.6645	0.6651
0.0372	5	570	0.6620	0.125	24.9697	0.6728	0.5674	0.6739	0.0	0.6722	0.6728
0.0372	6	684	0.6381	0.25	30.7010	0.6716	0.4617	0.6722	0.0	0.6710	0.6716
0.1652	7	798	0.6437	0.5	30.2020	0.6651	0.3994	0.6657	0.0	0.6645	0.6651
0.64	8.0	912	0.6202	1.0	43.3331	0.6881	0.4875	0.6887	0.0	0.6881	0.6881
0.4933	9.0	1026	0.6350	1.0	44.4736	0.6822	0.5892	0.6822	0.0	0.6822	0.6816
0.2369	10.0	1140	1.0023	1.0	47.0805	0.6091	0.5782	0.6085	0.0	0.6091	0.6091
0.1119	11.0	1254	2.2488	1.0	39.5203	0.6085	0.5847	0.6085	0.0	0.6091	0.6091
0.034	12.0	1368	2.0856	1.0	43.4524	0.6427	0.5816	0.6427	0.0	0.6436	0.6427

Framework versions

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