t5-3b_rte_sp0_ar0

This model is a fine-tuned version of t5-3b on the glue dataset. It achieves the following results on the evaluation set:

• Loss: 3.4268
• Accuracy: 0.8876

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: 4
• eval_batch_size: 8
• seed: 1
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 20
• training_steps: 750

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.7306	0.18	25	0.6913	0.5921
0.6717	0.36	50	0.4976	0.8339
0.3978	0.53	75	0.5226	0.8628
0.322	0.71	100	0.3902	0.8484
0.2958	0.89	125	0.3803	0.8881
0.2604	1.07	150	0.8628	0.8736
0.2011	1.25	175	0.7780	0.8953
0.263	1.42	200	2.1533	0.8881
0.2032	1.6	225	4.7955	0.8917
0.2536	1.78	250	1.7810	0.8989
0.1984	1.96	275	0.5119	0.8845
0.1495	2.14	300	0.5128	0.8845
0.1275	2.31	325	0.8602	0.8628
0.0955	2.49	350	1.3642	0.8773
0.3912	2.67	375	1.0186	0.8664
0.1108	2.85	400	2.1450	0.8592
0.0726	3.02	425	2.6801	0.8809
0.0937	3.2	450	5.2053	0.8736
1.0143	3.38	475	3.3979	0.8845
0.5754	3.56	500	4.2786	0.8989
0.2928	3.74	525	5.6543	0.8917
0.5633	3.91	550	6.7064	0.8845
1.0431	4.09	575	4.9205	0.8953
0.2839	4.27	600	4.2344	0.8809
0.5464	4.45	625	4.9598	0.8809
0.0031	4.63	650	5.3705	0.8881
0.5149	4.8	675	4.8105	0.8845
0.2702	4.98	700	6.9958	0.8953
0.7503	5.16	725	5.4360	0.8881
0.2639	5.34	750	5.4420	0.8917

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.1+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0