IndoT5-LoRA-R8-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: 0.5253

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
1.0213	1.0	188	0.9309
0.8595	2.0	376	0.8744
0.8909	3.0	564	0.8449
0.8118	4.0	752	0.8331
0.868	5.0	940	0.8022
0.7772	6.0	1128	0.7627
0.6771	7.0	1316	0.6054
0.6124	8.0	1504	0.5824
0.5767	9.0	1692	0.5603
0.5598	10.0	1880	0.5485
0.5429	11.0	2068	0.5430
0.5446	12.0	2256	0.5396
0.5809	13.0	2444	0.5374
0.5416	14.0	2632	0.5386
0.5573	15.0	2820	0.5324
0.5185	16.0	3008	0.5337
0.5278	17.0	3196	0.5284
0.5797	18.0	3384	0.5208
0.5117	19.0	3572	0.5229
0.5371	20.0	3760	0.5253

Framework versions

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