deepseek-8b-orpo-full

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>

deepseek-8b-orpo-full

This model is a fine-tuned version of deepseek-ai/deepseek-math-7b-base on the zfz1/my_preference_gsm8k_deepseek dataset. It achieves the following results on the evaluation set:

• Loss: 0.0797
• Rewards/chosen: -0.0089
• Rewards/rejected: -0.2196
• Rewards/accuracies: 1.0
• Rewards/margins: 0.2107
• Logps/rejected: -2.1960
• Logps/chosen: -0.0885
• Logits/rejected: 13.8906
• Logits/chosen: 8.5947
• Nll Loss: 0.0786
• Log Odds Ratio: -0.0204
• Log Odds Chosen: 5.3533

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 43
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 2

Training results

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1