IndoT5-LoRA-R16-LR0.0001-Ep20

This model is a fine-tuned version of Wikidepia/IndoT5-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5162

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
0.9753	1.0	188	0.8993
0.8341	2.0	376	0.8525
0.8659	3.0	564	0.8227
0.7751	4.0	752	0.7936
0.6818	5.0	940	0.5933
0.597	6.0	1128	0.5700
0.6174	7.0	1316	0.5574
0.5717	8.0	1504	0.5420
0.549	9.0	1692	0.5373
0.5388	10.0	1880	0.5337
0.5247	11.0	2068	0.5304
0.5271	12.0	2256	0.5259
0.5618	13.0	2444	0.5244
0.5242	14.0	2632	0.5216
0.5394	15.0	2820	0.5209
0.502	16.0	3008	0.5184
0.512	17.0	3196	0.5156
0.5617	18.0	3384	0.5115
0.495	19.0	3572	0.5119
0.5186	20.0	3760	0.5162

Framework versions

• PEFT 0.18.1
• Transformers 4.57.6
• Pytorch 2.10.0+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2