phi-3-mini-LoRA

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: 0.5586

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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
0.7527	0.1131	250	0.6002
0.5924	0.2262	500	0.5809
0.5811	0.3393	750	0.5759
0.5827	0.4524	1000	0.5717
0.5767	0.5655	1250	0.5704
0.5711	0.6787	1500	0.5678
0.5691	0.7918	1750	0.5672
0.5635	0.9049	2000	0.5654
0.5712	1.0180	2250	0.5650
0.5611	1.1311	2500	0.5647
0.555	1.2442	2750	0.5631
0.5505	1.3573	3000	0.5628
0.5657	1.4704	3250	0.5624
0.563	1.5835	3500	0.5617
0.5577	1.6966	3750	0.5614
0.5578	1.8098	4000	0.5603
0.5552	1.9229	4250	0.5604
0.5514	2.0360	4500	0.5600
0.5473	2.1491	4750	0.5603
0.5573	2.2622	5000	0.5596
0.5423	2.3753	5250	0.5599
0.5579	2.4884	5500	0.5595
0.5403	2.6015	5750	0.5591
0.5475	2.7146	6000	0.5593
0.5477	2.8277	6250	0.5590
0.5438	2.9408	6500	0.5586

Framework versions

• PEFT 0.12.0
• Transformers 4.43.1
• Pytorch 2.4.0a0+3bcc3cddb5.nv24.07
• Datasets 2.20.0
• Tokenizers 0.19.1