gemma-7b-dpo-full-mix1-beta-0.1-epoch-3

This model is a fine-tuned version of lewtun/gemma-7b-sft-full-deita-10k-v0 on the argilla/dpo-mix-7k dataset. It achieves the following results on the evaluation set:

• Loss: 0.5433
• Rewards/chosen: -3.1690
• Rewards/rejected: -5.2516
• Rewards/accuracies: 0.7188
• Rewards/margins: 2.0826
• Logps/rejected: -504.0652
• Logps/chosen: -485.0945
• Logits/rejected: 94.2728
• Logits/chosen: 100.2497

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: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.1371	1.9	100	0.5052	-2.9796	-5.1227	0.7708	2.1431	-502.7766	-483.2003	96.3785	102.3770

Framework versions

• Transformers 4.39.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.1