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: 0.6902
Rewards/chosen: -0.0213
Rewards/rejected: -0.0282
Rewards/accuracies: 0.4945
Rewards/margins: 0.0070
Logps/rejected: -28.6665
Logps/chosen: -23.4567
Logits/rejected: -2.8649
Logits/chosen: -2.8651

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: 4
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 8
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
0.6911	0.1	50	0.6909	0.0027	-0.0025	0.4967	0.0052	-28.5807	-23.3768	-2.8653	-2.8655
0.6916	0.2	100	0.6928	-0.0010	-0.0023	0.4571	0.0014	-28.5802	-23.3891	-2.8653	-2.8655
0.6924	0.29	150	0.6922	-0.0091	-0.0117	0.4681	0.0026	-28.6115	-23.4162	-2.8651	-2.8654
0.6941	0.39	200	0.6914	-0.0066	-0.0109	0.4879	0.0043	-28.6088	-23.4078	-2.8652	-2.8654
0.6942	0.49	250	0.6911	-0.0070	-0.0120	0.4791	0.0050	-28.6123	-23.4090	-2.8649	-2.8652
0.6909	0.59	300	0.6921	-0.0151	-0.0181	0.4593	0.0030	-28.6327	-23.4362	-2.8650	-2.8653
0.696	0.68	350	0.6903	-0.0140	-0.0207	0.5121	0.0067	-28.6414	-23.4326	-2.8651	-2.8653
0.6907	0.78	400	0.6904	-0.0153	-0.0217	0.4945	0.0064	-28.6448	-23.4369	-2.8649	-2.8652
0.6895	0.88	450	0.6898	-0.0157	-0.0232	0.4945	0.0075	-28.6497	-23.4380	-2.8649	-2.8652
0.6902	0.98	500	0.6892	-0.0192	-0.0282	0.5165	0.0090	-28.6665	-23.4500	-2.8650	-2.8652
0.6923	1.07	550	0.6893	-0.0196	-0.0282	0.5385	0.0086	-28.6663	-23.4511	-2.8649	-2.8652
0.6957	1.17	600	0.6897	-0.0210	-0.0288	0.5011	0.0078	-28.6684	-23.4560	-2.8649	-2.8652
0.6885	1.27	650	0.6897	-0.0173	-0.0251	0.5143	0.0078	-28.6560	-23.4436	-2.8650	-2.8653
0.6912	1.37	700	0.6906	-0.0207	-0.0268	0.4967	0.0061	-28.6617	-23.4548	-2.8650	-2.8652
0.6874	1.46	750	0.6903	-0.0216	-0.0282	0.4923	0.0065	-28.6663	-23.4580	-2.8650	-2.8652
0.6896	1.56	800	0.6877	-0.0180	-0.0298	0.5451	0.0119	-28.6719	-23.4457	-2.8649	-2.8651
0.6904	1.66	850	0.6905	-0.0217	-0.0279	0.4791	0.0062	-28.6655	-23.4582	-2.8649	-2.8651
0.6913	1.76	900	0.6902	-0.0213	-0.0282	0.4945	0.0070	-28.6665	-23.4567	-2.8649	-2.8651
0.6977	1.86	950	0.6902	-0.0213	-0.0282	0.4945	0.0070	-28.6665	-23.4567	-2.8649	-2.8651
0.6892	1.95	1000	0.6902	-0.0213	-0.0282	0.4945	0.0070	-28.6665	-23.4567	-2.8649	-2.8651

Framework versions

Transformers 4.38.2
Pytorch 2.0.0+cu117
Datasets 2.18.0
Tokenizers 0.15.2

tsavage68/mistralit2_1000_STEPS_1e8_rate_03_beta_DPO

mistralit2_1000_STEPS_1e8_rate_03_beta_DPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
mistralit2_1000_STEPS_1e8_rate_03_beta_DPO