CeLLaTe-ner-2class-pubmedbert-tapt-combData-tokenizer-adapted-baseline

This model is a fine-tuned version of Mardiyyah/CeLLaTe-tapt-pubmedbert-tokenizer-adapted-baseline-combinedData on the OTAR3088/CeLLaTe-ner-2class-iob_final dataset. It achieves the following results on the evaluation set:

• Loss: 0.0997
• Precision: 0.7821
• Recall: 0.7535
• Micro F1: 0.7675
• Weighted F1: 0.7678
• Macro F1: 0.7765
• Accuracy: 0.9845

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 3407
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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_ratio: 0.01
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	Micro F1	Weighted F1	Macro F1	Accuracy
0.2456	1.0	263	0.0655	0.5680	0.6957	0.6254	0.6265	0.6334	0.9776
0.0403	2.0	526	0.0568	0.7239	0.6795	0.7010	0.7009	0.6974	0.9822
0.0256	3.0	789	0.0579	0.7702	0.7234	0.7460	0.7461	0.7503	0.9837
0.0187	4.0	1052	0.0662	0.7530	0.7276	0.7401	0.7406	0.7524	0.9837
0.0143	5.0	1315	0.0653	0.7464	0.7096	0.7275	0.7275	0.7146	0.9833
0.0112	6.0	1578	0.0696	0.7356	0.7126	0.7239	0.7240	0.7135	0.9829
0.0083	7.0	1841	0.0747	0.7359	0.7589	0.7472	0.7476	0.7551	0.9844
0.0065	8.0	2104	0.0829	0.7665	0.7246	0.7450	0.7450	0.7471	0.9839
0.0056	9.0	2367	0.0845	0.7562	0.7462	0.7512	0.7513	0.7574	0.9841
0.0046	10.0	2630	0.0899	0.7752	0.7029	0.7373	0.7373	0.7402	0.9834
0.0039	11.0	2893	0.0848	0.7368	0.7661	0.7512	0.7521	0.7635	0.9839
0.0036	12.0	3156	0.0880	0.7762	0.7384	0.7569	0.7573	0.7663	0.9846
0.0029	13.0	3419	0.0942	0.7506	0.7619	0.7562	0.7564	0.7639	0.9845
0.0025	14.0	3682	0.0936	0.7637	0.7619	0.7628	0.7632	0.7730	0.9846
0.0023	15.0	3945	0.0970	0.7648	0.7625	0.7636	0.7639	0.7724	0.9845
0.0018	16.0	4208	0.0988	0.7717	0.7583	0.7649	0.7654	0.7750	0.9846
0.0017	17.0	4471	0.1000	0.7821	0.7535	0.7675	0.7678	0.7765	0.9845
0.0016	18.0	4734	0.1010	0.7648	0.7589	0.7618	0.7622	0.7713	0.9844
0.0014	19.0	4997	0.1044	0.7752	0.7529	0.7639	0.7642	0.7742	0.9844
0.0014	20.0	5260	0.1044	0.7719	0.7529	0.7623	0.7626	0.7729	0.9843

Framework versions

• Transformers 4.48.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.2
• Tokenizers 0.21.0