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: 0.6110
Rewards/chosen: -1.2720
Rewards/rejected: -1.7206
Rewards/accuracies: 0.6066
Rewards/margins: 0.4486
Logps/rejected: -34.3078
Logps/chosen: -27.6260
Logits/rejected: -2.8431
Logits/chosen: -2.8435

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: 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: 250

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6777	0.1	50	0.6740	-0.1496	-0.1942	0.5824	0.0446	-29.2197	-23.8845	-2.8632	-2.8635
0.6077	0.2	100	0.6364	-1.2703	-1.6253	0.5846	0.3550	-33.9902	-27.6202	-2.8384	-2.8387
0.4954	0.29	150	0.6401	-1.8310	-2.3459	0.5890	0.5149	-36.3920	-29.4892	-2.8362	-2.8366
0.5666	0.39	200	0.6112	-1.2659	-1.7103	0.6132	0.4444	-34.2733	-27.6055	-2.8434	-2.8438
0.6352	0.49	250	0.6110	-1.2720	-1.7206	0.6066	0.4486	-34.3078	-27.6260	-2.8431	-2.8435

Framework versions

Transformers 4.38.2
Pytorch 2.0.0+cu117
Datasets 2.18.0
Tokenizers 0.15.2

tsavage68/mistralit2_250_STEPS_1e7_rate_03_beta_DPO

mistralit2_250_STEPS_1e7_rate_03_beta_DPO

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

tsavage68/
mistralit2_250_STEPS_1e7_rate_03_beta_DPO