Google DeepMind’s Post

LLMs struggle to sound human when they speak because we are always technically reading their thoughts, and NOT what they are actually speaking or would have said. Directly out of the box. LLMs have no internal monologues. Unless we force it. Do REACT or monologue prompts and then it gets "space" / tokens to think. Let it sit and reflect on its thoughts. Let it think and know that it's thoughts are going to be spoke aloud. Better results.

Motion exploration: Infinite grid to fullscreen carousel using JS and GSAP FLIP. Images generated by Isaac Fayemi in Midjourney.

Want to run a powerful large language model on your machine? 🌟 GDE Vikram Tiwari dives into Gemma 2, Google's new open-weight LLM, exploring its capabilities, benchmarks, and how to get started running it locally. Check it out → https://goo.gle/4ey2Eqs

Spinning a web of code 🕷️ 🕸️ Code 👇 💻 spurs = 13; rings = 30; ang = linspace(0,2*rings,spurs*rings+1); r = logspace(-1,1,rings*spurs+1); h=voronoi(cospi(ang).*r, sinpi(ang).*r); delete(h(1)); text(0,0,'🕷','FontSize',40,'HorizontalAlignment','center'); axis([-10 10 -10 10],'square')

Because http://woc.space is difficult to pronounce, we decided to change the name to http://hoppa.space after some discussion today. “Hoppa” is Swedish for “jump” or “leap.” In different contexts, it can describe the action of jumping or convey a sense of excitement. In Dutch, “hoppa” is also commonly used, but it is more often an expression of emotion.

LLMs have become potent Agents, that can do a lot of our work by the way of text commands. The outputs of these agents have to be stored for our future reference. Copy pasting is the most troublesome, repetitive task. This post will show how Lua, LunarVim can automate the process

Temporal is an open source implementation of Durable Execution created by the originators of the abstraction. I am studying in it ......

MixSP: Space Decomposition for Sentence Embedding (ACL 2024 Finding) We introduce a novel embedding space decomposition method called "MixSP". In this paper, we challenge the common practice of treating STS scores as a continuous spectrum (0-5) to 2 ranges: "upper (0-4)" and "lower (4-5)" using "Mixture of Specialized Projectors." The experimental results demonstrate that MixSP outperforms existing sentence embedding methods on STS and zero-shot benchmarks. Paper: https://lnkd.in/dmqchyvt GitHub: https://lnkd.in/dy3pE_g7 Authors: Wuttikorn Ponwitayarat*, Peerat Limkonchotiwat*, Ekapol Chuangsuwanich, Sarana Nutanong Affiliations: VISTEC, Thailand and CU, Thailand.

Day 13 & 14 are about deep diving to understand Go’s Garbage Collection. Go’s garbage collection (GC) automatically manages memory, preventing memory leaks and optimizing performance by cleaning up unused objects. It’s designed to run in the background, so it doesn’t slow down your program. Go uses a mark-and-sweep method that is: - Marking: It identifies objects still in use. - Sweeping: It reclaims memory from unused objects. This process runs concurrently with your program, keeping performance smooth and minimising pauses. The Importance of GC: - Prevents Memory Leaks: GC eliminates manual memory management, reducing errors. - Optimized Performance: Go’s GC runs efficiently, freeing memory without halting the program. Optimizing Garbage Collection in Go: - Minimize Allocations: Avoid allocating memory in tight loops or frequently called functions. - Use Object Pools: Reuse objects with sync.Pool instead of creating new ones. - Tune GC with GOGC: Adjust GOGC to control GC frequency. The default is 100. Monitoring Garbage Collection - Profiling with pprof: Track memory usage and GC activity using Go’s pprof tool. - Adjusting GOGC: Change GOGC to make GC run more or less frequently based on your needs. Go’s GC automates memory cleanup, reducing leaks and optimizing performance. By understanding GC and following best practices, you can keep your program efficient, even at scale. #Golang #GoProgramming #SoftwareEngineering #ProgrammingTips #30DaysOfCode #30DaysOfCoding #LearningInPublic #OpenToWork #GoGC #GoLangTips #TechGrowth

📣 Day 18/30 of the challenge with 𝐓𝐮𝐫𝐧𝐢𝐧𝐠 𝐂𝐥𝐮𝐛 𝐨𝐟 𝐏𝐫𝐨𝐠𝐫𝐚𝐦𝐦𝐞𝐫𝐬 𝐍𝐈𝐓𝐑𝐑 🚀 Today’s task was the intriguing problem of “𝐊𝐭𝐡 𝐋𝐚𝐫𝐠𝐞𝐬𝐭 𝐄𝐥𝐞𝐦𝐞𝐧𝐭 𝐢𝐧 𝐚𝐧 𝐀𝐫𝐫𝐚𝐲.” 𝘓𝘦𝘵'𝘴 𝘥𝘪𝘷𝘦 𝘪𝘯𝘵𝘰 𝘪𝘵! 📥 𝐏𝐫𝐨𝐛𝐥𝐞𝐦 𝐒𝐭𝐚𝐭𝐞𝐦𝐞𝐧𝐭: Given an integer array nums and an integer k, return the kth largest element in the array. Note that it is the kth largest element in the sorted order, not the kth distinct element. Can you solve it without sorting?. ✅ Successfully solved this using the 𝐇𝐞𝐚𝐩 𝐃𝐚𝐭𝐚 𝐒𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞. Heap property: for every node, the value of its children is less than or equal to its own value. 💻 Resource: https://lnkd.in/gwJHfSsm Question: https://lnkd.in/gKYCuVNJ Keep up the great work!  🔸#30daysofcodewithTCP2024  🔸#TCPNITRR2024  🔸#codeutsavanitrr2024!  Let’s make this journey even more effective! 🌟