fine-tuning-dolphin-mistral-with-webglm-qa-with-lora_1

This model is a fine-tuned version of cognitivecomputations/dolphin-2.8-mistral-7b-v02 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2999

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: 700
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.7558	0.16	10	1.4842
1.4966	0.32	20	1.3367
1.2328	0.48	30	1.1282
0.9873	0.64	40	1.0817
0.9661	0.8	50	0.9967
0.8808	0.96	60	0.8844
0.7455	1.13	70	0.7337
0.6018	1.29	80	0.6164
0.4899	1.45	90	0.5440
0.4402	1.61	100	0.4971
0.4154	1.77	110	0.4555
0.4025	1.93	120	0.4238
0.3992	2.09	130	0.4007
0.3585	2.25	140	0.3862
0.3369	2.41	150	0.3666
0.3328	2.57	160	0.3537
0.3216	2.73	170	0.3423
0.2859	2.89	180	0.3303
0.2967	3.05	190	0.3211
0.2933	3.22	200	0.3114
0.2716	3.38	210	0.3097
0.255	3.54	220	0.3053
0.2731	3.7	230	0.2990
0.2729	3.86	240	0.2972
0.2701	4.02	250	0.3030
0.2558	4.18	260	0.3042
0.2612	4.34	270	0.3301
0.3048	4.5	280	0.4564
0.5437	4.66	290	0.7938
1.5888	4.82	300	1.5418
0.6588	4.98	310	0.4630
0.5345	5.14	320	0.9088
1.1475	5.31	330	1.6381
1.6442	5.47	340	2.0495
2.2517	5.63	350	1.7558
0.9492	5.79	360	0.5187
0.3727	5.95	370	0.3763
0.3139	6.11	380	0.3376
0.2896	6.27	390	0.3195
0.283	6.43	400	0.3106
0.2646	6.59	410	0.3105
0.2674	6.75	420	0.3256
0.3482	6.91	430	0.4016
0.4193	7.07	440	0.6300
0.7397	7.23	450	1.0617
1.1954	7.4	460	1.6157
1.6177	7.56	470	1.8019
1.2996	7.72	480	0.9151
0.6605	7.88	490	0.5433
0.416	8.04	500	0.4012
0.3412	8.2	510	0.3685
0.3322	8.36	520	0.3928
0.3516	8.52	530	0.3641
0.3406	8.68	540	0.4061
0.3772	8.84	550	0.4145
0.3695	9.0	560	0.5453
0.5824	9.16	570	0.7332
0.5139	9.32	580	0.4839
0.3798	9.49	590	0.3758
0.319	9.65	600	0.3438
0.3082	9.81	610	0.3301
0.3017	9.97	620	0.3225
0.2862	10.13	630	0.3156
0.2586	10.29	640	0.3109
0.2878	10.45	650	0.3082
0.2766	10.61	660	0.3056
0.2834	10.77	670	0.3042
0.2513	10.93	680	0.3020
0.2762	11.09	690	0.3007
0.28	11.25	700	0.2999

Framework versions

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