I noticed that in the quantizer.py, you have a init_embed_ function which use data to init the embedding weight. In a distributed training environment, each rank has their own data, which leads to ...
A Streamlit-based application that enables natural language querying of a PostgreSQL database using OpenAI's GPT models for SQL generation and pgvector for semantic search capabilities.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback