mobilebert_sa_GLUE_Experiment_logit_kd_qqp

This model is a fine-tuned version of google/mobilebert-uncased on the GLUE QQP dataset. It achieves the following results on the evaluation set:

• Loss: 0.6021
• Accuracy: 0.8081
• F1: 0.7357
• Combined Score: 0.7719

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: 128
• eval_batch_size: 128
• seed: 10
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Combined Score
0.9285	1.0	2843	0.8489	0.7517	0.6660	0.7089
0.7919	2.0	5686	0.7104	0.7735	0.6483	0.7109
0.6916	3.0	8529	0.6615	0.7900	0.6994	0.7447
0.6382	4.0	11372	0.6606	0.7899	0.6842	0.7370
0.6003	5.0	14215	0.6277	0.7988	0.7181	0.7585
0.5696	6.0	17058	0.6174	0.7980	0.7058	0.7519
0.5434	7.0	19901	0.6062	0.8036	0.7266	0.7651
0.5186	8.0	22744	0.6182	0.7995	0.7232	0.7614
0.4968	9.0	25587	0.6087	0.8052	0.7257	0.7654
0.4758	10.0	28430	0.6035	0.8073	0.7359	0.7716
0.456	11.0	31273	0.6021	0.8081	0.7357	0.7719
0.4361	12.0	34116	0.6137	0.8070	0.7366	0.7718
0.4186	13.0	36959	0.6282	0.8076	0.7416	0.7746
0.4009	14.0	39802	0.6183	0.8093	0.7445	0.7769
0.3846	15.0	42645	0.6196	0.8057	0.7483	0.7770
0.3699	16.0	45488	0.6381	0.8122	0.7426	0.7774

Framework versions

• Transformers 4.26.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.9.0
• Tokenizers 0.13.2