The best *code embedding* model in the market right now was just released: Qodo-Embed-1 — There are two flavors: A lite model with 1.5B parameters and a medium model with 7B parameters (Hugging Face links below). If you want to index a large codebase (supports 10M+ lines of code),... this is the model you want. 1. Index your repositories 2. Ask anything (including test and code generation) The models are optimized to answer natural language questions or code-to-code questions. The video here shows the model indexing 90 repositories (!!!!!) and letting the user ask questions about them. The simplest way to use the model is through the Qodo Gen AI extension in Visual Studio Code, Cursor, or JetBrains (see link below).show more

Here is a link to the Visual Studio Code extension: And here is the link to the JetBrains extension: Here is the Hugging Face link to the Lite model (1.5B): And here is the link to the Medium model (7B):

Join me as I put several Machine Learning models head-to-head to see which one can beat the market and deliver strong returns. In this free Substack post I share several models that deliver better returns with much lower drawdown compared to Buy-and-Hold approach.

More behind-the-scenes details: relevant data, synthetic data, and fitting training:

A Tel Aviv company.

That's quite a claim

How do you make that embedding available if using Claude 3.7 for coding? Or do you switch for coding to Qodo also?

Is it related to the AlphaCodium paper in some way? How do you compare to solutions that pass the code as context like agentic chat or sourcegraph search / Cody? thanks

Entonces ya está resuelto el problema de software?

Nos va a tomar mucho más que un modelo para code embedding para resolver el problema del software.

Very impressive. 👍

@cursor_ai Planning to add this as an embedding option?