phi-3-mini-QLoRA

This model is a fine-tuned version of microsoft/Phi-3-mini-4k-instruct on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5741

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
1.1422	0.1810	100	0.6625
0.6238	0.3619	200	0.6002
0.5932	0.5429	300	0.5906
0.5926	0.7238	400	0.5860
0.5794	0.9048	500	0.5834
0.5868	1.0857	600	0.5815
0.5711	1.2667	700	0.5796
0.5729	1.4476	800	0.5785
0.5858	1.6286	900	0.5772
0.5732	1.8095	1000	0.5763
0.5736	1.9905	1100	0.5756
0.5638	2.1715	1200	0.5754
0.5769	2.3524	1300	0.5746
0.5668	2.5334	1400	0.5745
0.5675	2.7143	1500	0.5742
0.5693	2.8953	1600	0.5741

Framework versions

• PEFT 0.10.0
• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1