hugpadhugpad

alignment-handbook / zephyr-7b-dpo-full

Text Generation7.2B1103Updated Jan 10, 2024
Hugging Face

Downloads

110

Likes

3

Last updated

Jan 10, 2024

License

apache-2.0

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the HuggingFaceH4/ultrafeedbackbinarized dataset. It achieves the following results on the evaluation set: - Loss: 0.5042 - Rewards/chosen: -1.0500 - Rewards/rejected: -2.0480 - Rewards/accuracies: …

Tasks & library

  • text generation
  • transformers

Tags

  • transformers
  • tensorboard
  • mistral
  • text-generation
  • alignment-handbook
  • generated_from_trainer

Technical specifications

Parameters7.2B
Librarytransformers
LanguagesNot available
Base modelalignment-handbook/zephyr-7b-sft-full
Licenseapache-2.0