Mike Tamir, PhD’s Post

Checked context, you can plug the encoder i'm working on into any LLM and it should work, AKA you can crunch context sizes with minimal effect on the quality itself!

I really think that machines from HTB are way more developed then THM, but both are very useful for studying, I am now migrating to Hackthebox, and in this EASY room I had some hard time but could make it trough watching videos and following mr. Bishal Ray -#GxbNt writeup. This way I could understand how to harden shell, and understand that if we can see, have patience, and keep on searching for more resources online, we understand how it works.

Understanding of building block concepts allow you to build knowledge on top. Today, I chat about what a #vector embedding is and aligning your prompt style to your use case in #GenAI applications. #btt

JEP-461 is ready as a preview in JDK 22. Looks like an interesting one. A good number of intermediate operations for Stream API are on the way!

#expensive #awkard and #dangerous that so many testers optimize for speed. Wrong priority especially in a world where test design/infra means tests are easily parallelizable. Human time and cost is far far more expensive and embarassing and career-limitiing vs testing infra time and cost 🤡 #AI #LLM Overall Quality vs Speed vs Pricing Info Claude 3.5 Sonnet wins for #serious testing FWIW: for the #testing infrastructure I'm working on, I prioritize Quality first, pricing second, and performance/speed third.

This guide covers the core set of principles you can apply to improve latency across a wide variety of LLM-related used cases. https://lnkd.in/dcKC67SQ

🙌 What is a Test Algorithm? Algorithms serve as the circuitry that governs the functionality of the program. A test algorithm is essentially a specialized set of instructions designed to carefully evaluate the performance, accuracy, and reliability of another algorithm. It acts as a quality assurance checkpoint, ensuring that the underlying algorithm is working properly. Why are test algorithms crucial? 👇 ✔ Performance Evaluation ✔ Accuracy Verification ✔ Reliability Assessment 📌 Want to delve deeper into test algorithms? Share your thoughts or questions in the comments below! #softwaredevelopment #algorithms #testing #qualityassurance

Reterival Augment Generation is a difficult pattern to crack, and anyone who claims that a Nocode tool can solve all problems or that it only takes three lines to develop a chat system falls into the same category as the get rich scheme. This article discusses 12 locations where the RAG system might fail, and how to tweak and measure the failure. I don't agree with the answers because they depend, but it's good to know you can try these. Great article by Wenqi Glantz. Thanks https://lnkd.in/gbizx-wr

Common errors in use of terms : Out-of-box solution - means a solution ready made and available for implementation; { standard solution if you so will } Out-of-the-box - means a radical, novel, as yet untested solution.

Neat pricing tool to compare cost of many LLMs, all on the same page: https://lnkd.in/e3h2GgR5 We built out an in-house tool for viewing costs across our system & pull out most costly calls.... after they've already been made. How do other people monitor LLM calls & manage cost?