bert-base-uncased_08112024T144127

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: 0.4242
• F1: 0.8849
• 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	1.7911	0.1661	0.0000
No log	2.0	173	1.6895	0.2558	0.0000
No log	2.9942	259	1.5467	0.4336	0.0000
No log	4.0	346	1.3716	0.5012	0.0000
No log	4.9942	432	1.1818	0.5459	0.0000
1.5117	6.0	519	1.0336	0.5938	0.0000
1.5117	6.9942	605	0.9389	0.6309	1e-05
1.5117	8.0	692	0.8480	0.6802	0.0000
1.5117	8.9942	778	0.7481	0.7288	0.0000
1.5117	10.0	865	0.6824	0.7561	0.0000
1.5117	10.9942	951	0.6213	0.7867	0.0000
0.7682	12.0	1038	0.5781	0.8039	0.0000
0.7682	12.9942	1124	0.5184	0.8345	0.0000
0.7682	14.0	1211	0.4854	0.8489	0.0000
0.7682	14.9942	1297	0.4815	0.8559	0.0000
0.7682	16.0	1384	0.4422	0.8704	0.0000
0.7682	16.9942	1470	0.4422	0.8761	6e-06
0.305	18.0	1557	0.4368	0.8791	0.0000
0.305	18.9942	1643	0.4242	0.8849	0.0000
0.305	20.0	1730	0.4483	0.8829	0.0000
0.305	20.9942	1816	0.4539	0.8841	0.0000
0.305	22.0	1903	0.4521	0.8862	0.0000
0.305	22.9942	1989	0.4450	0.8896	0.0000
0.1014	24.0	2076	0.4603	0.8874	0.0000
0.1014	24.9942	2162	0.4750	0.8864	0.0000
0.1014	26.0	2249	0.4711	0.8887	7e-07
0.1014	26.9942	2335	0.4756	0.8879	4e-07
0.1014	28.0	2422	0.4691	0.8883	2e-07
0.0521	28.9942	2508	0.4694	0.8883	0.0
0.0521	29.8266	2580	0.4699	0.8883	0.0

Framework versions

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