The community member is experiencing an issue where they are getting a "RuntimeError: context has already been set" error, likely due to the use of the 'fork' method instead of the 'spawn' method for GPU usage in the llama_index library. Another community member suggests that the issue may depend on the specific LLM or other features being used, as they have not seen the 'fork' or 'spawn' methods used in the llama_index codebase. The original poster is using llama-cpp for the LLM, BGE for embedding, and torch, and plans to check if torch is setting the method explicitly.
I tried but couldn't get it work. Most likely, because I need to use 'spawn' method for GPU and llama_index might be setting it to 'fork' somewhere in the code? Getting this error: RuntimeError: context has already been set
