fine-tuning-Phi2-with-webglm-qa-with-lora_8

This model is a fine-tuned version of microsoft/phi-2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0935

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: 5
• total_train_batch_size: 10
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 60
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
7.1823	0.31	20	6.1082
4.0	0.63	40	0.9863
0.7159	0.94	60	0.6293
0.4994	1.26	80	0.4239
0.3187	1.57	100	0.3044
0.251	1.89	120	0.2567
0.2189	2.2	140	0.2206
0.1869	2.52	160	0.2000
0.1741	2.83	180	0.1781
0.1439	3.14	200	0.1638
0.1543	3.46	220	0.1550
0.1428	3.77	240	0.1455
0.127	4.09	260	0.1394
0.1206	4.4	280	0.1314
0.1206	4.72	300	0.1298
0.1162	5.03	320	0.1246
0.109	5.35	340	0.1235
0.1088	5.66	360	0.1190
0.1062	5.97	380	0.1157
0.0938	6.29	400	0.1146
0.0945	6.6	420	0.1133
0.1012	6.92	440	0.1105
0.0881	7.23	460	0.1109
0.0897	7.55	480	0.1091
0.0837	7.86	500	0.1060
0.0899	8.18	520	0.1051
0.0803	8.49	540	0.1041
0.0792	8.81	560	0.1021
0.0885	9.12	580	0.1000
0.0844	9.43	600	0.1004
0.0704	9.75	620	0.0992
0.0681	10.06	640	0.0994
0.0727	10.38	660	0.0977
0.0712	10.69	680	0.0970
0.073	11.01	700	0.0971
0.0683	11.32	720	0.0974
0.0682	11.64	740	0.0964
0.0716	11.95	760	0.0962
0.0645	12.26	780	0.0948
0.0662	12.58	800	0.0947
0.0677	12.89	820	0.0947
0.0626	13.21	840	0.0953
0.0628	13.52	860	0.0946
0.0642	13.84	880	0.0937
0.0641	14.15	900	0.0939
0.0587	14.47	920	0.0939
0.0664	14.78	940	0.0933
0.061	15.09	960	0.0931
0.0596	15.41	980	0.0934
0.0646	15.72	1000	0.0935

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.0.0
• Datasets 2.15.0
• Tokenizers 0.15.0