bert-base-uncased_12112024T103207

This model is a fine-tuned version of google-bert/bert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4280
• F1: 0.8755
• Learning Rate: 0.0

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 600
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Rate
No log	0.9942	86	1.7552	0.1858	0.0000
No log	2.0	173	1.6269	0.3187	0.0000
No log	2.9942	259	1.4885	0.4438	0.0000
No log	4.0	346	1.3478	0.4980	0.0000
No log	4.9942	432	1.1903	0.5445	0.0000
1.5065	6.0	519	1.0219	0.5810	0.0000
1.5065	6.9942	605	0.9065	0.6140	1e-05
1.5065	8.0	692	0.7955	0.6526	0.0000
1.5065	8.9942	778	0.6876	0.7032	0.0000
1.5065	10.0	865	0.6171	0.7536	0.0000
1.5065	10.9942	951	0.5734	0.7612	0.0000
0.7171	12.0	1038	0.4960	0.8147	0.0000
0.7171	12.9942	1124	0.4820	0.8358	0.0000
0.7171	14.0	1211	0.4557	0.8445	0.0000
0.7171	14.9942	1297	0.4596	0.8524	0.0000
0.7171	16.0	1384	0.4299	0.8651	0.0000
0.7171	16.9942	1470	0.4426	0.8671	6e-06
0.2382	18.0	1557	0.4280	0.8755	0.0000
0.2382	18.9942	1643	0.4517	0.8728	0.0000
0.2382	20.0	1730	0.4473	0.8761	0.0000
0.2382	20.9942	1816	0.4599	0.8798	0.0000
0.2382	22.0	1903	0.4927	0.8777	0.0000
0.2382	22.9942	1989	0.4768	0.8819	0.0000
0.0713	24.0	2076	0.4970	0.8808	0.0000
0.0713	24.9942	2162	0.5031	0.8808	0.0000
0.0713	26.0	2249	0.4807	0.8845	7e-07
0.0713	26.9942	2335	0.4959	0.8825	4e-07
0.0713	28.0	2422	0.5034	0.8818	2e-07
0.0344	28.9942	2508	0.5037	0.8818	0.0
0.0344	29.8266	2580	0.5037	0.8824	0.0

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.19.1