Mistral-Nemo-12B-Instruct-SAA-Half

This model is a fine-tuned version of mistralai/Mistral-Nemo-Instruct-2407 on the bct_non_cot_dpo_500 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5932
• Rewards/chosen: -0.0554
• Rewards/rejected: -0.0900
• Rewards/accuracies: 0.7800
• Rewards/margins: 0.0346
• Logps/rejected: -0.9001
• Logps/chosen: -0.5538
• Logits/rejected: -2.3124
• Logits/chosen: -2.2862
• Sft Loss: 0.0563
• Odds Ratio Loss: 5.3690

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
• gradient_accumulation_steps: 8
• total_train_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: 3.0

Training results

Framework versions

• PEFT 0.12.0
• Transformers 4.45.2
• Pytorch 2.3.0
• Datasets 2.19.0
• Tokenizers 0.20.0