Mistral-7B-Instruct-v0.2-MI-2e-5

• Loss: 1.5846
• Rewards/chosen: -0.8695
• Rewards/rejected: -0.9117
• Rewards/accuracies: 0.5559
• Rewards/margins: 0.0422
• Logps/rejected: -0.9117
• Logps/chosen: -0.8695
• Logits/rejected: -2.8388
• Logits/chosen: -2.8404

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

• learning_rate: 2e-05
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 16
• total_train_batch_size: 128
• total_eval_batch_size: 16
• 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

Framework versions

• Transformers 4.42.0
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.19.1