td2

TD2 - ESGI

Ahmed Ennaifer & Sarra Chabane Chaouche

This model is a fine-tuned version of almanach/camembert-base on the None dataset. It achieves the following results on the evaluation set of train/test:

• Loss: 0.0128
• Precision: 0.0
• Recall: 0.0
• F1: 0.0
• Accuracy: 0.9983

The model got an accuracy of 0.998 on the seperate eval dataset : test_fr

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• 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
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	160	0.0455	0.0	0.0	0.0	0.9975
No log	2.0	320	0.0292	0.0	0.0	0.0	0.9981
No log	3.0	480	0.0224	0.0	0.0	0.0	0.9981
0.0837	4.0	640	0.0188	0.0	0.0	0.0	0.9981
0.0837	5.0	800	0.0166	0.0	0.0	0.0	0.9979
0.0837	6.0	960	0.0148	0.0	0.0	0.0	0.9981
0.0182	7.0	1120	0.0139	0.0	0.0	0.0	0.9982
0.0182	8.0	1280	0.0133	0.0	0.0	0.0	0.9981
0.0182	9.0	1440	0.0129	0.0	0.0	0.0	0.9982
0.0122	10.0	1600	0.0128	0.0	0.0	0.0	0.9983

Framework versions

• Transformers 4.50.0
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.1