𝗧𝗵𝗲 𝗕𝗲𝗻𝗲𝗳𝗶𝗍𝘀 𝗢𝗳 𝗖𝗼𝗺𝗯𝗶𝗻𝗶𝗻𝗴 𝗟𝗼𝗰𝗮𝗹 𝗮𝗻𝗱 𝗛𝗼𝘀𝗍𝗲𝗱 𝗟𝗟𝗠

𝗧𝗵𝗲 𝗕𝗲𝗻𝗲𝗳𝗶𝗍𝘀 𝗢𝗳 𝗖𝗼𝗺𝗯𝗶𝗻𝗶𝗻𝗴 𝗟𝗼𝗰𝗮𝗹 𝗮𝗻𝗱 𝗛𝗼𝘀𝗍𝗲𝗱 𝗟𝗟𝗠 I am working on a large project called gas-fakes. It allows local execution, continuous integration, and containerization of native Apps Script code.

• We are liberating Apps Script code
• We are making it more accessible

My project has 4399 methods and 10,500 parity tests. I am now using AI to help with coding work.

• I use a local model to do the heavy work
• I use a hosted model for planning and decision-making

You can use this technique to save time and money.

• You can use a local model to reduce token costs
• You can use a hosted model for high-level planning

Here's how it works:

• The hosted model determines what needs to be done
• The local model executes the task
• The hosted model integrates the result into the final response

To get started, you need to set up a local model orchestrator.

• You can use oMLX on a Mac
• You can use other orchestrators on other devices

You can control the system behavior via environment variables.

• You can verify that everything is working
• You can check the oMlx dashboard

Source: https://dev.to/brucemcpherson/combining-local-and-hosted-llm-to-minimize-token-cost-o69 Optional learning community: https://t.me/GyaanSetuAi