Modelpublic

tsavage68/UTI_M2_1000steps_1e7rate_SFT

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face

0likes10downloads

Model Card

UTIM21000steps1e7rateSFT

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.2 on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 1.1047

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: 1e-07
—trainbatchsize: 2
—evalbatchsize: 1
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 4
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: cosine
—lrschedulerwarmup_steps: 100
—training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss
2.4021	0.3333	25	2.3941
2.3235	0.6667	50	2.2471
2.0863	1.0	75	1.9386
1.6662	1.3333	100	1.5791
1.2956	1.6667	125	1.2544
1.214	2.0	150	1.2116
1.202	2.3333	175	1.1861
1.1813	2.6667	200	1.1668
1.1696	3.0	225	1.1528
1.1052	3.3333	250	1.1412
1.0614	3.6667	275	1.1329
1.1106	4.0	300	1.1271
1.1019	4.3333	325	1.1228
1.0691	4.6667	350	1.1212
1.0947	5.0	375	1.1153
1.0689	5.3333	400	1.1134
1.0598	5.6667	425	1.1116
1.0459	6.0	450	1.1111
1.0518	6.3333	475	1.1097
1.045	6.6667	500	1.1092
1.0658	7.0	525	1.1066
1.0706	7.3333	550	1.1067
1.0514	7.6667	575	1.1057
1.0412	8.0	600	1.1063
1.0455	8.3333	625	1.1052
0.9657	8.6667	650	1.1057
1.1015	9.0	675	1.1052
1.0294	9.3333	700	1.1051
1.0399	9.6667	725	1.1052
1.1125	10.0	750	1.1047
1.0219	10.3333	775	1.1046
0.9862	10.6667	800	1.1048
1.0682	11.0	825	1.1049
1.0587	11.3333	850	1.1049
1.0217	11.6667	875	1.1051
1.0547	12.0	900	1.1047
1.0047	12.3333	925	1.1047
1.021	12.6667	950	1.1047
1.0528	13.0	975	1.1047
1.0385	13.3333	1000	1.1047

Framework versions

—Transformers 4.41.2
—Pytorch 2.0.0+cu117
—Datasets 2.19.2
—Tokenizers 0.19.1