evacun-lemmatization-vanilla-second-split

This model is a fine-tuned version of bowphs/evacun2025-1 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2926
• Exact Match: 0.8811

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match
0.0534	5.9999	27542	0.1986	0.8633
0.0399	6.9999	32132	0.2140	0.8587
0.0275	8.0	36723	0.2305	0.86
0.0204	8.9999	41313	0.2508	0.8654
0.0154	9.9998	45903	0.2628	0.8728
0.0124	11.0000	50494	0.2812	0.8745
0.0107	11.9999	55084	0.2865	0.8734
0.0088	12.9998	59674	0.2926	0.8811
0.0067	13.9999	64265	0.3054	0.8648
0.0055	14.9999	68855	0.3151	0.8668
0.0054	16.0	73446	0.3202	0.8737
0.0042	16.9999	78036	0.3277	0.8695
0.0041	17.9998	82626	0.3282	0.8737

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.20.0