bert-tiny-mlm-finetuned-imdb

This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.4487

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: 3e-05
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• num_epochs: 200

Training results

Training Loss	Epoch	Step	Validation Loss
4.1774	1.04	500	3.7705
4.041	2.09	1000	3.7196
3.9982	3.13	1500	3.6826
3.9614	4.18	2000	3.6543
3.9274	5.22	2500	3.6438
3.9089	6.26	3000	3.6294
3.8929	7.31	3500	3.6217
3.873	8.35	4000	3.6083
3.8659	9.39	4500	3.5900
3.8484	10.44	5000	3.5791
3.8261	11.48	5500	3.5731
3.8228	12.53	6000	3.5579
3.8098	13.57	6500	3.5576
3.8028	14.61	7000	3.5532
3.7881	15.66	7500	3.5440
3.7829	16.7	8000	3.5440
3.7727	17.75	8500	3.5372
3.7648	18.79	9000	3.5248
3.7504	19.83	9500	3.5223
3.7487	20.88	10000	3.5212
3.7497	21.92	10500	3.5166
3.7344	22.96	11000	3.5103
3.7339	24.01	11500	3.5052
3.722	25.05	12000	3.5067
3.7188	26.1	12500	3.4941
3.7127	27.14	13000	3.4951
3.7113	28.18	13500	3.4904
3.7042	29.23	14000	3.4813
3.7011	30.27	14500	3.4805
3.6936	31.32	15000	3.4886
3.6889	32.36	15500	3.4825
3.6771	33.4	16000	3.4785
3.6753	34.45	16500	3.4819
3.6743	35.49	17000	3.4744
3.6686	36.53	17500	3.4658
3.669	37.58	18000	3.4607
3.6623	38.62	18500	3.4688
3.6648	39.67	19000	3.4676
3.6574	40.71	19500	3.4581
3.652	41.75	20000	3.4601
3.6506	42.8	20500	3.4630
3.6466	43.84	21000	3.4530
3.637	44.89	21500	3.4507
3.6428	45.93	22000	3.4557
3.6408	46.97	22500	3.4483
3.6368	48.02	23000	3.4505
3.6322	49.06	23500	3.4494
3.6256	50.1	24000	3.4487

Framework versions

• Transformers 4.25.1
• Pytorch 1.12.1+cu113
• Datasets 2.7.1
• Tokenizers 0.13.2