d0ae661759c2cce6d7b28a1aafc469d6

This model is a fine-tuned version of facebook/opt-125m on the nyu-mll/glue [mnli] dataset. It achieves the following results on the evaluation set:

• Loss: 0.7854
• Data Size: 1.0
• Epoch Runtime: 851.5614
• Accuracy: 0.7658
• F1 Macro: 0.7662
• Rouge1: 0.7660
• Rouge2: 0.0
• Rougel: 0.7661
• Rougelsum: 0.7661

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	1.3477	0	7.9227	0.3216	0.2659	0.3216	0.0	0.3214	0.3215
1.0258	1	12271	0.8069	0.0078	14.4409	0.6539	0.6516	0.6539	0.0	0.6541	0.6541
0.7755	2	24542	0.7102	0.0156	20.9267	0.7003	0.6951	0.7003	0.0	0.7004	0.7004
0.6898	3	36813	0.6649	0.0312	33.9091	0.7259	0.7233	0.7259	0.0	0.7260	0.7258
0.6516	4	49084	0.6227	0.0625	60.9415	0.7445	0.7442	0.7442	0.0	0.7448	0.7445
0.5937	5	61355	0.6299	0.125	115.6915	0.7392	0.7395	0.7391	0.0	0.7394	0.7392
0.5922	6	73626	0.6302	0.25	217.2937	0.7414	0.7410	0.7413	0.0	0.7414	0.7413
0.5301	7	85897	0.5712	0.5	422.6086	0.7671	0.7665	0.7671	0.0	0.7670	0.7673
0.5094	8.0	98168	0.5676	1.0	829.6305	0.7794	0.7789	0.7792	0.0	0.7794	0.7794
0.4175	9.0	110439	0.5926	1.0	832.6565	0.7773	0.7757	0.7774	0.0	0.7773	0.7774
0.3659	10.0	122710	0.6209	1.0	844.8599	0.7758	0.7737	0.7755	0.0	0.7759	0.7757
0.337	11.0	134981	0.7233	1.0	853.2940	0.7650	0.7653	0.7649	0.0	0.7651	0.7650
0.2426	12.0	147252	0.7854	1.0	851.5614	0.7658	0.7662	0.7660	0.0	0.7661	0.7661

Framework versions

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