README.md

metadata

license: mit
base_model: microsoft/Phi-3-mini-4k-instruct
tags:
  - generated_from_trainer
model-index:
  - name: PHI30512HMAB2
    results: []

PHI30512HMAB2

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: 0.0706

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_steps: 80
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
5.4852	0.09	10	5.4372
5.4761	0.18	20	5.3249
5.1807	0.27	30	4.6301
4.0639	0.36	40	3.0706
2.2191	0.45	50	1.1829
0.7606	0.54	60	0.3950
0.2685	0.63	70	0.1655
0.1453	0.73	80	0.1335
0.1187	0.82	90	0.1252
0.1302	0.91	100	0.1241
0.1114	1.0	110	0.1113
0.1009	1.09	120	0.0967
0.089	1.18	130	0.0971
0.1047	1.27	140	0.0844
0.0839	1.36	150	0.0811
0.0876	1.45	160	0.0815
0.0769	1.54	170	0.0813
0.081	1.63	180	0.0765
0.0673	1.72	190	0.0762
0.0789	1.81	200	0.0767
0.0643	1.9	210	0.0739
0.0715	1.99	220	0.0748
0.0643	2.08	230	0.0729
0.0626	2.18	240	0.0722
0.0565	2.27	250	0.0722
0.0594	2.36	260	0.0722
0.0629	2.45	270	0.0717
0.0594	2.54	280	0.0719
0.0627	2.63	290	0.0712
0.0582	2.72	300	0.0705
0.0659	2.81	310	0.0706
0.0603	2.9	320	0.0705
0.0649	2.99	330	0.0706

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.14.0