gemma-3-4b-it-arkey_emails-qlora

This model is a fine-tuned version of google/gemma-3-4b-it on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.5329

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: 5e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 64
• total_train_batch_size: 128
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss
No log	1.0	5	3.9168
No log	2.0	10	3.5567
No log	3.0	15	3.2439
No log	4.0	20	3.0028
No log	5.0	25	2.8257
No log	6.0	30	2.6974
No log	7.0	35	2.6045
No log	8.0	40	2.5319
No log	9.0	45	2.4673
No log	10.0	50	2.4065
No log	11.0	55	2.3477
No log	12.0	60	2.2918
No log	13.0	65	2.2379
No log	14.0	70	2.1864
No log	15.0	75	2.1368
No log	16.0	80	2.0882
No log	17.0	85	2.0429
No log	18.0	90	1.9999
No log	19.0	95	1.9580
No log	20.0	100	1.9175
No log	21.0	105	1.8778
No log	22.0	110	1.8372
No log	23.0	115	1.7964
No log	24.0	120	1.7642
No log	25.0	125	1.7354
No log	26.0	130	1.7083
No log	27.0	135	1.6831
No log	28.0	140	1.6598
No log	29.0	145	1.6386
No log	30.0	150	1.6197
No log	31.0	155	1.6029
No log	32.0	160	1.5880
No log	33.0	165	1.5750
No log	34.0	170	1.5638
No log	35.0	175	1.5543
No log	36.0	180	1.5466
No log	37.0	185	1.5407
No log	38.0	190	1.5365
No log	39.0	195	1.5339
No log	40.0	200	1.5329

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.1.2
• Datasets 3.5.1
• Tokenizers 0.21.1