bert_base_train_book

This model is a fine-tuned version of distilbert-base-uncased on the gokulsrinivasagan/processed_book_corpus_cleaned dataset. It achieves the following results on the evaluation set:

• Loss: 1.0775
• Accuracy: 0.7531

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: 0.0001
• train_batch_size: 96
• eval_batch_size: 96
• seed: 10
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10000
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
5.6277	0.4215	10000	5.4679	0.1648
5.5308	0.8431	20000	5.3921	0.1656
5.4819	1.2646	30000	5.3559	0.1668
5.4576	1.6861	40000	5.3327	0.1669
5.434	2.1077	50000	5.3193	0.1671
5.423	2.5292	60000	5.3064	0.1676
5.4078	2.9507	70000	5.3011	0.1670
5.3996	3.3723	80000	5.2891	0.1675
5.3864	3.7938	90000	5.2806	0.1672
5.3883	4.2153	100000	5.2894	0.1641
5.3743	4.6369	110000	5.2662	0.1678
5.3614	5.0584	120000	5.2495	0.1677
2.7786	5.4799	130000	2.4132	0.5314
2.191	5.9014	140000	1.8931	0.6135
1.997	6.3230	150000	1.7234	0.6414
1.8894	6.7445	160000	1.6208	0.6582
1.801	7.1660	170000	1.5466	0.6709
1.7429	7.5876	180000	1.4959	0.6795
1.6988	8.0091	190000	1.4521	0.6867
1.6587	8.4306	200000	1.4160	0.6930
1.6247	8.8522	210000	1.3884	0.6977
1.5996	9.2737	220000	1.3623	0.7023
1.5686	9.6952	230000	1.3387	0.7062
1.5445	10.1168	240000	1.3201	0.7099
1.5316	10.5383	250000	1.3002	0.7128
1.51	10.9598	260000	1.2850	0.7156
1.4938	11.3814	270000	1.2728	0.7178
1.4864	11.8029	280000	1.2574	0.7205
1.4641	12.2244	290000	1.2453	0.7228
1.4549	12.6460	300000	1.2324	0.7250
1.4394	13.0675	310000	1.2212	0.7270
1.4298	13.4890	320000	1.2135	0.7284
1.4227	13.9106	330000	1.2044	0.7299
1.414	14.3321	340000	1.1946	0.7319
1.4028	14.7536	350000	1.1855	0.7333
1.3929	15.1751	360000	1.1794	0.7344
1.3863	15.5967	370000	1.1696	0.7360
1.3762	16.0182	380000	1.1627	0.7372
1.3697	16.4397	390000	1.1562	0.7387
1.36	16.8613	400000	1.1513	0.7395
1.3566	17.2828	410000	1.1425	0.7411
1.3482	17.7043	420000	1.1388	0.7417
1.3398	18.1259	430000	1.1331	0.7430
1.3332	18.5474	440000	1.1295	0.7436
1.3316	18.9689	450000	1.1221	0.7448
1.3235	19.3905	460000	1.1177	0.7457
1.321	19.8120	470000	1.1127	0.7464
1.3123	20.2335	480000	1.1087	0.7474
1.3069	20.6551	490000	1.1046	0.7480
1.3016	21.0766	500000	1.0994	0.7486
1.2977	21.4981	510000	1.0952	0.7497
1.2929	21.9197	520000	1.0932	0.7500
1.2924	22.3412	530000	1.0899	0.7505
1.2862	22.7627	540000	1.0887	0.7510
1.2853	23.1843	550000	1.0847	0.7517
1.2827	23.6058	560000	1.0813	0.7523
1.2787	24.0273	570000	1.0805	0.7524
1.276	24.4488	580000	1.0765	0.7532
1.2732	24.8704	590000	1.0770	0.7530

Framework versions

• Transformers 4.46.1
• Pytorch 2.2.0+cu121
• Datasets 3.1.0
• Tokenizers 0.20.1