selective-pairrm-33079692-mt2

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7212
• Rewards/chosen: -2.3879
• Rewards/rejected: -2.5044
• Rewards/accuracies: 0.5742
• Rewards/margins: 0.1164
• Logps/rejected: -817.7779
• Logps/chosen: -781.7959
• Logits/rejected: -3.0082
• Logits/chosen: -3.0229

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: 5e-07
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.3678	0.32	100	0.8166	-3.5115	-3.6302	0.5703	0.1186	-930.3584	-894.1545	-3.2211	-3.2330
0.5115	0.64	200	0.7558	-2.9722	-3.0863	0.5781	0.1141	-875.9759	-840.2244	-2.8707	-2.8897
0.6156	0.96	300	0.7196	-2.3861	-2.5020	0.5703	0.1159	-817.5390	-781.6083	-3.0086	-3.0233

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2
• Datasets 2.14.6
• Tokenizers 0.15.0