zephyr-dpo-timedial-selfgen-mix2

This model is a fine-tuned version of EllieS/zephyr-dpo-timedial-selfgen on the EllieS/timedial_dpo dataset. It achieves the following results on the evaluation set:

• Logits/chosen: -2.8807
• Logits/rejected: -2.8679
• Logps/chosen: -6.8403
• Logps/rejected: -132.2824
• Loss: 0.2588
• Rewards/accuracies: 1.0
• Rewards/chosen: 0.2087
• Rewards/margins: 1.2287
• Rewards/rejected: -1.0200

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-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• 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_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Logits/chosen	Logits/rejected	Logps/chosen	Logps/rejected	Validation Loss	Rewards/accuracies	Rewards/chosen	Rewards/margins	Rewards/rejected
0.4825	0.35	100	-2.9715	-2.9751	-4.6766	-60.5070	0.4632	1.0	0.2303	0.5325	-0.3022
0.2786	0.69	200	-2.8807	-2.8679	-6.8403	-132.2824	0.2588	1.0	0.2087	1.2287	-1.0200

Framework versions

• PEFT 0.7.1
• Transformers 4.39.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2