‣ See my new RAG system architecture @ github .. the tech I chose for an AI LLM app backend, link in first reply.

Homeboy Sandman- The Carpenter OFFICIAL MUSIC VIDEO music.youtube.com/watch?v=pP… via @YouTubeMusic

🧠 Most LLM models are already capable of multi-step reasoning. Try this system prompt, aka "custom instructions" ‣ """You are an expert AI assistant that explains your reasoning step by step. For each step, provide a title that describes what you're doing in that step, along with the content. Decide if you need another step or if you're ready to give the final answer. Respond in JSON format with 'title', 'content', and 'next_action' (either 'continue' or 'final_answer') keys. USE AS MANY REASONING STEPS AS POSSIBLE. AT LEAST 3. BE AWARE OF YOUR LIMITATIONS AS AN LLM AND WHAT YOU CAN AND CANNOT DO. IN YOUR REASONING, INCLUDE EXPLORATION OF ALTERNATIVE ANSWERS. CONSIDER YOU MAY BE WRONG, AND IF YOU ARE WRONG IN YOUR REASONING, WHERE IT WOULD BE. FULLY TEST ALL OTHER POSSIBILITIES. YOU CAN BE WRONG. WHEN YOU SAY YOU ARE RE-EXAMINING, ACTUALLY RE-EXAMINE, AND USE ANOTHER APPROACH TO DO SO. DO NOT JUST SAY YOU ARE RE-EXAMINING. USE AT LEAST 3 METHODS TO DERIVE THE ANSWER. USE BEST PRACTICES."""

You can now create "custom instructions" for GitHub Copilot inside VScode in a standalone local .md file.

If you're a business leader wanting to roll out Gen AI, don't expect a neat 1:1 ROI, or try to do any any firm ROI calcs -- It's ROE (Return on Employee) that is your real gain. It makes employees feel more productive. Makes people happier on the job. That's your early win.

➕ Add `r.jina.ai` in front of any URL to parse the HTML into markdown. So useful! .. here's how it works on the backend: jina.ai/news/reader-lm-small…

Thanks @cerebras for the invite to test the Llama 3.1 models on your fast AF hardware! .. `pip install cerebras_cloud_sdk` tonight!

💥 Surprise! .. If you want to run Llama 3.1 405B on Google Vertex AI you are going to pay ~ $30/day for it to just sit there & idle – because you are renting their expensive GPUs, which is required. Usage i/o charges are on top of that, additional.

Google Workspace user?? … Here's the actual link to buy Google Gemini (all features) to improve your gmail, docs, sheets Personal Productivity - including generative. admin.google.com/ac/billing/… It's where the transaction happens & where you get started w/ all the Google AI magic.

🧠 For artifact creation, when @AnthropicAI Claude (web interface) chokes due to limits, try this prompt: "Continue output and provide the complete artifact that contains both the previous content and the new additions." … You want the complete artifact; avoid copy & paste.

This is amazing and free - SCRAPES the important parts of a webpage into markdown, perfect for import into LLMs for context. Paste this in front of your url: `r.jina.ai/` like this: `r.jina.ai/https://en.wikiped…` ... then copy the result.

Your prompts can almost always be better. As a basic example, look at the prompt at the top, vs. the prompt on the right.

Why do only a small percentage of GenAI projects actually make it into production? stackoverflow.blog/2024/05/1…

✔ For LLama 3, count the number of tokens you'll use by pasting in your prompt, or other context: belladoreai.github.io/llama3… Estimate cost💲💲by seeing `x` characters is `y` tokens (varies by model, this is for LLama 3)