This model is a fine-tuned version of tsavage68/Summary4500_L3_100steps_1e6rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.7020
Rewards/chosen: 0.0041
Rewards/rejected: -0.0059
Rewards/accuracies: 0.5240
Rewards/margins: 0.0099
Logps/rejected: -133.2091
Logps/chosen: -84.1817
Logits/rejected: -1.0990
Logits/chosen: -1.0689

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-08
train_batch_size: 1
eval_batch_size: 1
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6884	0.0112	50	0.7086	0.0095	0.0109	0.5040	-0.0013	-133.1756	-84.1708	-1.0984	-1.0681
0.7459	0.0224	100	0.7035	0.0204	0.0127	0.5220	0.0077	-133.1719	-84.1490	-1.0983	-1.0679
0.6447	0.0336	150	0.7056	-0.0034	-0.0106	0.5100	0.0072	-133.2185	-84.1965	-1.0983	-1.0682
0.6548	0.0448	200	0.6858	0.0309	-0.0169	0.5580	0.0478	-133.2311	-84.1280	-1.0987	-1.0684
0.7161	0.0559	250	0.6867	0.0348	-0.0097	0.5380	0.0446	-133.2168	-84.1201	-1.0990	-1.0690
0.6022	0.0671	300	0.6971	0.0079	-0.0133	0.5140	0.0212	-133.2239	-84.1739	-1.0989	-1.0689
0.6858	0.0783	350	0.7002	0.0098	-0.0060	0.5320	0.0158	-133.2093	-84.1702	-1.0988	-1.0687
0.9044	0.0895	400	0.6925	0.0134	-0.0174	0.5500	0.0308	-133.2322	-84.1629	-1.0991	-1.0691
0.7801	0.1007	450	0.6901	0.0173	-0.0219	0.5480	0.0392	-133.2412	-84.1552	-1.0986	-1.0687
0.6132	0.1119	500	0.6968	0.0225	0.0018	0.5220	0.0207	-133.1937	-84.1448	-1.0990	-1.0689
0.7548	0.1231	550	0.6830	0.0105	-0.0372	0.5700	0.0477	-133.2716	-84.1687	-1.0988	-1.0687
0.6306	0.1343	600	0.6850	0.0305	-0.0153	0.5520	0.0458	-133.2280	-84.1289	-1.0991	-1.0690
0.7374	0.1454	650	0.6811	0.0135	-0.0388	0.5600	0.0522	-133.2749	-84.1629	-1.0992	-1.0691
0.6897	0.1566	700	0.6801	0.0231	-0.0330	0.5640	0.0561	-133.2632	-84.1436	-1.0988	-1.0687
0.7106	0.1678	750	0.6939	0.0137	-0.0135	0.5340	0.0272	-133.2244	-84.1625	-1.0991	-1.0690
0.6409	0.1790	800	0.6842	0.0163	-0.0302	0.5700	0.0465	-133.2576	-84.1571	-1.0990	-1.0688
0.6427	0.1902	850	0.7002	0.0013	-0.0130	0.5240	0.0143	-133.2233	-84.1871	-1.0990	-1.0689
0.7061	0.2014	900	0.7020	0.0041	-0.0059	0.5240	0.0099	-133.2091	-84.1817	-1.0990	-1.0689
0.6772	0.2126	950	0.7020	0.0041	-0.0059	0.5240	0.0099	-133.2091	-84.1817	-1.0990	-1.0689
0.6709	0.2238	1000	0.7020	0.0041	-0.0059	0.5240	0.0099	-133.2091	-84.1817	-1.0990	-1.0689

Framework versions

Transformers 4.42.4
Pytorch 2.0.0+cu117
Datasets 2.20.0
Tokenizers 0.19.1

tsavage68/Summary4500_L3_1000steps_1e8rate_05beta_CSFTDPO

Hyponatremia_L3_1000steps_1e8rate_05beta_CSFTDPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
Summary4500_L3_1000steps_1e8rate_05beta_CSFTDPO