zephyr-7b-gemma-dpo

This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-gemma-sft-v0.1 on the argilla/dpo-mix-7k dataset. It achieves the following results on the evaluation set:

• Loss: 0.4658
• Rewards/chosen: -4.1134
• Rewards/rejected: -5.8777
• Rewards/accuracies: 0.7292
• Rewards/margins: 1.7644
• Logps/rejected: -479.6258
• Logps/chosen: -445.9359
• Logits/rejected: 85.7594
• Logits/chosen: 91.4868

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: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• 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: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.1497	1.8957	100	0.4608	-4.1286	-5.8457	0.7292	1.7171	-478.9854	-446.2400	85.7808	91.4893

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1