1bc41893e06ff09fbcbc02f2671a0903

This model is a fine-tuned version of distilbert/distilbert-base-cased-distilled-squad on the nyu-mll/glue [mnli] dataset. It achieves the following results on the evaluation set:

• Loss: 0.7452
• Data Size: 1.0
• Epoch Runtime: 326.0347
• Accuracy: 0.7804
• F1 Macro: 0.7800
• Rouge1: 0.7804
• Rouge2: 0.0
• Rougel: 0.7805
• Rougelsum: 0.7807

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	1.0998	0	3.0460	0.3544	0.1802	0.3543	0.0	0.3544	0.3543
1.0561	1	12271	0.8954	0.0078	5.8002	0.5978	0.5982	0.5979	0.0	0.5978	0.5980
0.8687	2	24542	0.8337	0.0156	8.3566	0.6505	0.6436	0.6504	0.0	0.6507	0.6508
0.7605	3	36813	0.7366	0.0312	13.1327	0.6849	0.6830	0.6850	0.0	0.6855	0.6849
0.7307	4	49084	0.6743	0.0625	23.5567	0.7226	0.7214	0.7226	0.0	0.7227	0.7228
0.6336	5	61355	0.6211	0.125	42.7324	0.7402	0.7389	0.7404	0.0	0.7407	0.7403
0.6121	6	73626	0.6075	0.25	81.4882	0.7492	0.7496	0.7492	0.0	0.7495	0.7494
0.5114	7	85897	0.5819	0.5	158.2857	0.7662	0.7640	0.7660	0.0	0.7662	0.7663
0.4736	8.0	98168	0.5647	1.0	317.9532	0.7746	0.7750	0.7745	0.0	0.7746	0.7747
0.4045	9.0	110439	0.6022	1.0	325.5091	0.7788	0.7779	0.7786	0.0	0.7790	0.7790
0.3278	10.0	122710	0.6238	1.0	325.6244	0.7761	0.7749	0.7759	0.0	0.7761	0.7765
0.3058	11.0	134981	0.7399	1.0	328.1703	0.7730	0.7731	0.7729	0.0	0.7733	0.7732
0.2274	12.0	147252	0.7452	1.0	326.0347	0.7804	0.7800	0.7804	0.0	0.7805	0.7807

Framework versions

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