IndoT5-LoRA-R16-LR0.0003-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: 0.4517

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: 0.0003
• 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
0.8889	1.0	188	0.8337
0.5949	2.0	376	0.5920
0.5928	3.0	564	0.5431
0.5278	4.0	752	0.5376
0.592	5.0	940	0.5176
0.5311	6.0	1128	0.5185
0.5608	7.0	1316	0.5151
0.4822	8.0	1504	0.4613
0.4491	9.0	1692	0.4542
0.434	10.0	1880	0.4504
0.4212	11.0	2068	0.4532
0.4184	12.0	2256	0.4378
0.4511	13.0	2444	0.4379
0.4175	14.0	2632	0.4419
0.427	15.0	2820	0.4517

Framework versions

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