RAG From Scratch: Feedback + self-reflection Our RAG From Scratch video series walks through impt RAG concepts in short / focused videos w/ code. This is the final part in our series, focusing on self-reflection + feedback to improve RAG systems. Problem: RAG systems
LLMS
-
Haiku vs Opus: Cost Comparison for AI Model Usage
By
–
If you use Haiku, it's quite cheap, probably under 50 cents for a book. Opus is much more expensive.
-
Claude 3 excels at summarizing long books but struggles with fact-checking
By
–
"Talk to your documents" is a big use case for AI. This paper tested how good AI was at working with long, new books (like The Fourth Wing). Claude 3 does very well in SUMMARIZING big books. But none of the models do a good job FACT CHECKING claims about books made by others.
-
Creating LLM Embeddings Plugins: Documentation and Examples
By
–
There's documentation on creating a new embeddings plugin here https://
llm.datasette.io/en/stable/embe
ddings/writing-plugins.html
… They are pretty simple though, so best approach is to look at one of the existing embeddings plugins and copy from that – here's one I wrote the other day https://
github.com/simonw/llm-nom
ic-api-embed/blob/main/llm_nomic_api_embed.py
… -
AI Models Getting Better Make Code Simpler
By
–
Pretty much, with some added fanciness here and there. As the models get better, the code gets simpler. Older versions of this were far more complex. I'd guess in a year, this could be done with just one or two prompts.
-
As Models Improve, Code Becomes Simpler
By
–
As models get better, code gets simpler 🙂 Take a look at some of the older versions, you'll see what I mean.
-
Building RAG Q&A Systems with LangChain.js and Azure
By
–
LangChain.js + Azure This talk from JavaScript Dev Day highlights how to build a RAG-based Q&A system using YouTube video transcripts. Uses open-source models and @ollama , and then transitions to Azure for production https://
youtube.com/watch?v=nYXSKs
8qMY8
… -
AI-Generated Books Now More Engaging Than Previous Versions
By
–
Shockingly good. Past versions (built with GPT-4) made sense but weren't fun to read. This actually produces interesting books.
-
GPT-Author Update: Claude 3 Powers Better Novel Generation
By
–
This is an update to the original gpt-author I built last year. It now uses Claude 3 for better writing and dialogue, and produces much, much more interesting novels. And the new Claude models are so much better that I was able to make the code much simpler than before!
-
Claude-Author: Constrained Agent for Coherent Novel Writing
By
–
Claude-Author is a constrained agent — meaning its behavior is highly-controlled, leading to better results than open-ended agents. It's a chain of many Claude 3 Haiku/Opus and Stable Diffusion API calls that work together to write a coherent novel.