myshell-ai / JetMoE

Reaching LLaMA2 Performance with 0.1M Dollars
Apache License 2.0
937 stars 72 forks source link

Parameter mapping #11

Open takgto opened 3 weeks ago

takgto commented 3 weeks ago

Could you please provide a mapping between the parameters of the jetmoe model (model.layes.{}....) and the parameters of the Hugging Face GPT model (transformer.h.{}..... ) ? I am very interested in using the jetmoe model and would like to use it with litgpt. To do that, we need to know the mapping to convert jetmoe huggingface into gpt model.