Hyponatremia_M2_400steps_1e8rate_03beta_DPO

This model is a fine-tuned version of tsavage68/Hyponatremia_M2_150steps_1e6rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6276
• Rewards/chosen: 0.0345
• Rewards/rejected: -0.1027
• Rewards/accuracies: 0.9600
• Rewards/margins: 0.1372
• Logps/rejected: -71.2238
• Logps/chosen: -36.6121
• Logits/rejected: -2.2455
• Logits/chosen: -2.2404

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: 2
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• training_steps: 400

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.691	0.2667	50	0.6959	0.0081	0.0125	0.4400	-0.0044	-70.8398	-36.7002	-2.2478	-2.2427
0.684	0.5333	100	0.6850	0.0066	-0.0109	0.6200	0.0175	-70.9177	-36.7050	-2.2462	-2.2410
0.6496	0.8	150	0.6433	0.0267	-0.0772	0.9000	0.1039	-71.1387	-36.6380	-2.2462	-2.2412
0.6234	1.0667	200	0.6319	0.0320	-0.0962	0.9700	0.1281	-71.2020	-36.6205	-2.2465	-2.2415
0.6262	1.3333	250	0.6336	0.0321	-0.0927	0.9500	0.1248	-71.1904	-36.6199	-2.2462	-2.2411
0.6215	1.6	300	0.6288	0.0386	-0.0965	0.9600	0.1352	-71.2033	-36.5983	-2.2455	-2.2404
0.6129	1.8667	350	0.6276	0.0345	-0.1027	0.9600	0.1372	-71.2238	-36.6121	-2.2455	-2.2404
0.6301	2.1333	400	0.6276	0.0345	-0.1027	0.9600	0.1372	-71.2238	-36.6121	-2.2455	-2.2404

Framework versions

• Transformers 4.42.3
• Pytorch 2.0.0+cu117
• Datasets 2.20.0
• Tokenizers 0.19.1