Na_L3_100steps_1e6rate_03beta_cSFTDPO

This model is a fine-tuned version of tsavage68/Na_L3_100steps_1e6rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0000
• Rewards/chosen: 1.5010
• Rewards/rejected: -10.3716
• Rewards/accuracies: 1.0
• Rewards/margins: 11.8726
• Logps/rejected: -76.0737
• Logps/chosen: -19.8868
• Logits/rejected: -0.9605
• Logits/chosen: -0.8913

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-06
• 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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.0001	0.2667	50	0.0001	1.2406	-8.7274	1.0	9.9680	-70.5931	-20.7550	-0.9609	-0.8923
0.0	0.5333	100	0.0000	1.5010	-10.3716	1.0	11.8726	-76.0737	-19.8868	-0.9605	-0.8913

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1