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: 0.0018
Rewards/chosen: 0.6548
Rewards/rejected: -9.1653
Rewards/accuracies: 0.9980
Rewards/margins: 9.8200
Logps/rejected: -151.5279
Logps/chosen: -82.8803
Logits/rejected: -1.1014
Logits/chosen: -1.0667

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
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
0.7013	0.0112	50	0.5626	0.0456	-0.2692	0.8000	0.3149	-133.7358	-84.0985	-1.0991	-1.0689
0.0899	0.0224	100	0.1139	0.1758	-2.1072	0.9980	2.2831	-137.4118	-83.8381	-1.1001	-1.0687
0.0007	0.0336	150	0.0084	0.3555	-5.4656	0.9980	5.8211	-144.1285	-83.4787	-1.1015	-1.0681
0.0002	0.0448	200	0.0037	0.4541	-6.9717	0.9980	7.4258	-147.1408	-83.2816	-1.1017	-1.0678
0.0002	0.0559	250	0.0028	0.5004	-7.6120	0.9980	8.1124	-148.4213	-83.1889	-1.1014	-1.0671
0.0	0.0671	300	0.0024	0.5292	-7.9130	0.9980	8.4422	-149.0233	-83.1313	-1.1011	-1.0669
0.0002	0.0783	350	0.0023	0.5504	-8.2153	0.9980	8.7657	-149.6280	-83.0890	-1.1010	-1.0665
0.0	0.0895	400	0.0021	0.5876	-8.5585	0.9980	9.1460	-150.3143	-83.0146	-1.1008	-1.0663
0.0	0.1007	450	0.0020	0.6154	-8.7473	0.9980	9.3626	-150.6919	-82.9590	-1.1011	-1.0665
0.0	0.1119	500	0.0019	0.6370	-8.8365	0.9980	9.4735	-150.8704	-82.9158	-1.1010	-1.0664
0.0	0.1231	550	0.0019	0.6457	-8.9971	0.9980	9.6429	-151.1916	-82.8983	-1.1008	-1.0662
0.0	0.1343	600	0.0018	0.6663	-9.0854	0.9980	9.7517	-151.3682	-82.8572	-1.1016	-1.0669
0.0	0.1454	650	0.0018	0.6239	-9.1522	0.9980	9.7760	-151.5017	-82.9421	-1.1006	-1.0658
0.0	0.1566	700	0.0018	0.6305	-9.1452	0.9980	9.7757	-151.4877	-82.9288	-1.1008	-1.0660
0.0012	0.1678	750	0.0018	0.6289	-9.1809	0.9980	9.8098	-151.5591	-82.9320	-1.1015	-1.0668
0.0	0.1790	800	0.0018	0.6367	-9.1807	0.9980	9.8174	-151.5587	-82.9164	-1.1008	-1.0660
0.0001	0.1902	850	0.0018	0.6608	-9.1943	0.9980	9.8551	-151.5860	-82.8683	-1.1015	-1.0667
0.0	0.2014	900	0.0018	0.6591	-9.1599	0.9980	9.8189	-151.5170	-82.8717	-1.1014	-1.0667
0.0	0.2126	950	0.0018	0.6596	-9.1677	0.9980	9.8273	-151.5327	-82.8705	-1.1014	-1.0667
0.0	0.2238	1000	0.0018	0.6548	-9.1653	0.9980	9.8200	-151.5279	-82.8803	-1.1014	-1.0667

Framework versions

Transformers 4.42.4
Pytorch 2.0.0+cu117
Datasets 2.20.0
Tokenizers 0.19.1

tsavage68/Summary4500_L3_1000steps_1e7rate_05beta_CSFTDPO

Hyponatremia_L3_1000steps_1e7rate_05beta_CSFTDPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
Summary4500_L3_1000steps_1e7rate_05beta_CSFTDPO