embc25_finetuned_fr

This model is a fine-tuned version of Kyungjin-Kim/mmc_roberta_500000_fr on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.4235
• Accuracy: 0.6202
• Precision: 0.6656
• Recall: 0.7790
• F1: 0.7179

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: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Use 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: 500
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.6276	0.64	500	0.9535	0.6236	0.6299	0.9529	0.7585
0.5567	1.2790	1000	0.6283	0.5258	0.7097	0.3986	0.5104
0.5278	1.9190	1500	0.8207	0.6022	0.6882	0.6558	0.6716
0.4862	2.5581	2000	0.8486	0.6079	0.7010	0.6413	0.6698
0.4116	3.1971	2500	0.8388	0.5652	0.6950	0.5326	0.6031
0.3825	3.8371	3000	0.6555	0.5292	0.7195	0.3949	0.5099
0.34	4.4762	3500	1.2660	0.6079	0.6822	0.6884	0.6853
0.2856	5.1152	4000	1.2518	0.6034	0.6780	0.6866	0.6823
0.2663	5.7552	4500	1.4022	0.6124	0.6832	0.6993	0.6911
0.2493	6.3942	5000	1.5075	0.5966	0.6764	0.6703	0.6733
0.2036	7.0333	5500	1.6113	0.6112	0.6846	0.6920	0.6883
0.2028	7.6733	6000	2.0145	0.6157	0.6705	0.7482	0.7072
0.1691	8.3123	6500	1.7540	0.6022	0.6807	0.6757	0.6782
0.1593	8.9523	7000	2.0229	0.6079	0.6712	0.7210	0.6952
0.1486	9.5914	7500	2.3382	0.6169	0.6667	0.7645	0.7122

Framework versions

• Transformers 4.48.1
• Pytorch 2.3.1
• Datasets 3.2.0
• Tokenizers 0.21.0