Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

ANTHROPIC 🔥: Claude will soon receive a new file-based memory upgrade, offering users the option to choose between Memory Files and Classic memory. > Organized notes Claude writes as you chat and reads when they're relevant. Browse and edit them anytime. This feature appears to be a new iteration... show more

🚨 AI News | TestingCatalog

71,489 subscribers

419,772 Aufrufe • vor 2 Monaten •via X (Twitter)

Nachrichten & Politik Bildung Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

ANTHROPIC 🔥: A new Managed Projects feature has been spotted in testing on Claude. > Claude takes on tasks and keeps the project organized. > A project is a home for one stream of work. Sessions share memory and instructions so context carries forward, and Claude runs more autonomously. Projects only create and manage cloud sessions. This feature may be built on top of Claude Managed Agents, where each project gets a dedicated cloud environment so Claude can execute periodic tasks and refine project context via "Dreams". It could also be a successor to Conway, which is set to be removed this Friday internally.

ANTHROPIC 🔥: A new Managed Projects feature has been spotted in testing on Claude. > Claude takes on tasks and keeps the project organized. > A project is a home for one stream of work. Sessions share memory and instructions so context carries forward, and Claude runs more autonomously. Projects only create and manage cloud sessions. This feature may be built on top of Claude Managed Agents, where each project gets a dedicated cloud environment so Claude can execute periodic tasks and refine project context via "Dreams". It could also be a successor to Conway, which is set to be removed this Friday internally.

🚨 AI News | TestingCatalog

61,827 Aufrufe • vor 13 Tagen

THIS GUY TURNS NOTES, DOCUMENTS, AND IDEAS INTO AN AI SECOND BRAIN the system stores all the context of your work and helps claude not forget anything how to build one yourself: install obsidian connect claude code keep your projects, notes, and sessions in one vault link similar ideas together use the graph to discover new connections as a result, claude gets memory of your projects and understands the context without you constantly having to explain everything the more you work, the smarter the system gets don’t collect notes build connections between them instead

THIS GUY TURNS NOTES, DOCUMENTS, AND IDEAS INTO AN AI SECOND BRAIN the system stores all the context of your work and helps claude not forget anything how to build one yourself: install obsidian connect claude code keep your projects, notes, and sessions in one vault link similar ideas together use the graph to discover new connections as a result, claude gets memory of your projects and understands the context without you constantly having to explain everything the more you work, the smarter the system gets don’t collect notes build connections between them instead

Marvin

24,954 Aufrufe • vor 1 Monat

New feature in Claude Code 2.1.14 just dropped! You can now search and install plugins from the marketplaces installed in your current Claude Code session. This is huge if you’re building plugins on top of Claude Code’s marketplace layer (Skills, Agents, Hooks, etc). How it works: - Run /plugin - The official Claude marketplace is installed by default - Use the search bar to find the plugin you want - Select one or multiple plugins with space, then press i to install - Go to the Installed tab to browse and enable them With the exponential growth of Skills and Agent-based components running in the CLI, improving plugin discoverability is a big win. Pretty sure more marketplace-related features are coming

New feature in Claude Code 2.1.14 just dropped! You can now search and install plugins from the marketplaces installed in your current Claude Code session. This is huge if you’re building plugins on top of Claude Code’s marketplace layer (Skills, Agents, Hooks, etc). How it works: - Run /plugin - The official Claude marketplace is installed by default - Use the search bar to find the plugin you want - Select one or multiple plugins with space, then press i to install - Go to the Installed tab to browse and enable them With the exponential growth of Skills and Agent-based components running in the CLI, improving plugin discoverability is a big win. Pretty sure more marketplace-related features are coming

Daniel San

40,994 Aufrufe • vor 6 Monaten

BREAKING 🚨: Microsoft released Copilot Appearance to selected users in US, UK and Canada as a part of Copilot Labs. "Copilot Appearance is an experiment that gives you a new, visual way to chat with Copilot, powered by real-time expressions, voice, and conversational memory."

BREAKING 🚨: Microsoft released Copilot Appearance to selected users in US, UK and Canada as a part of Copilot Labs. "Copilot Appearance is an experiment that gives you a new, visual way to chat with Copilot, powered by real-time expressions, voice, and conversational memory."

🚨 AI News | TestingCatalog

30,476 Aufrufe • vor 1 Jahr

Memories are the previous learnings and context. Developers build up state and context as they do work, which is what makes a more tenured developer more effective than a brand new one with a similar skillset. It would be both inefficient and painful to relearn information from scratch around code structure, architecture, etc each time you needed to do a new task. Memories solve for this. As you do work with Cascade, it can automatically choose to “remember” pieces of information that it learns as Memories, and for any later work, it can choose to pull from this memory bank instead of trying to relearn that information from scratch. You also can manually prompt Cascade to remember parts of conversations as Memories and can manually go in and edit Memories post-fact. Here’s a developer asking Cascade to save some knowledge as a Memory:

Windsurf

12,356 Aufrufe • vor 1 Jahr

Love and Deepspace | Rerun Event Preview The 5-Star Rate UP Pool [Twilight Serenity] Limited-Time Rerun will start soon! "Then, you're not allowed to change your mind even after a hundred years. Or a thousand." 💫Event Duration: From 05:00 on Dec. 11 to 04:59 on Dec. 18 (Server Time) 💫The 5-Star Rate UP Pool [Twilight Serenity] Limited-Time Rerun Event 1. During the event, make a wish with [Deepspace Wish] or [Time Wish: Limited] to participate in the wish event. The drop rate of the event-limited 5-Star Memory [Rafayel: Fireworks Vow] will go up drastically. 2. After the event ends, this limited 5-Star Memory will not be obtainable through other means and will not enter the permanent Wish Pool: Xspace Echo. 3. All Rerun Wish Pools share one pity system. A 5-Star Memory is guaranteed within a specific attempt of wishes. If the 5-Star Memory you have obtained from the Rerun Wish Pool is not the event-limited Memory, you will obtain the event-limited Memory the next time you obtain a 5-Star Memory. The pity count from the last Rerun Wish Pool can be applied to this Rerun Wish Pool, and the pity count in this Rerun Wish Pool will also be applied to the upcoming Rerun Wish Pool. *You can read more about the event on the in-game rules page. 🎁New Packs During the event, the Rerun event-exclusive [Flamebloom Pack] series, which includes [Time Wish: Limited] and other materials, will be available in Shop. Notes: 1. [Time Wish: Limited] can be used in 5-Star Memory Wish Pool Rerun and will be used first when you make a wish. 2. After the event ends, [Time Wish: Limited] will automatically convert to Empyrean Wish. ——— 🪐Official Discord: #LoveandDeepspace #Rafayel

Love and Deepspace | Rerun Event Preview The 5-Star Rate UP Pool [Twilight Serenity] Limited-Time Rerun will start soon! "Then, you're not allowed to change your mind even after a hundred years. Or a thousand." 💫Event Duration: From 05:00 on Dec. 11 to 04:59 on Dec. 18 (Server Time) 💫The 5-Star Rate UP Pool [Twilight Serenity] Limited-Time Rerun Event 1. During the event, make a wish with [Deepspace Wish] or [Time Wish: Limited] to participate in the wish event. The drop rate of the event-limited 5-Star Memory [Rafayel: Fireworks Vow] will go up drastically. 2. After the event ends, this limited 5-Star Memory will not be obtainable through other means and will not enter the permanent Wish Pool: Xspace Echo. 3. All Rerun Wish Pools share one pity system. A 5-Star Memory is guaranteed within a specific attempt of wishes. If the 5-Star Memory you have obtained from the Rerun Wish Pool is not the event-limited Memory, you will obtain the event-limited Memory the next time you obtain a 5-Star Memory. The pity count from the last Rerun Wish Pool can be applied to this Rerun Wish Pool, and the pity count in this Rerun Wish Pool will also be applied to the upcoming Rerun Wish Pool. *You can read more about the event on the in-game rules page. 🎁New Packs During the event, the Rerun event-exclusive [Flamebloom Pack] series, which includes [Time Wish: Limited] and other materials, will be available in Shop. Notes: 1. [Time Wish: Limited] can be used in 5-Star Memory Wish Pool Rerun and will be used first when you make a wish. 2. After the event ends, [Time Wish: Limited] will automatically convert to Empyrean Wish. ——— 🪐Official Discord: #LoveandDeepspace #Rafayel

Love and Deepspace

304,066 Aufrufe • vor 7 Monaten

HE BUILT A 100% PRIVATE SECOND BRAIN IN OBSIDIAN TO AUTOMATE HIS RESEARCH USING HERMES AGENT AND NOTEBOOKLM He visualizes his entire knowledge graph on-premise without paying for cloud subscriptions By connecting Hermes Agent and NotebookLM, developers can index hundreds of documents and generate content locally Four components of this local knowledge architecture: 1. Memory layer - set up an Obsidian PARA vault to store markdown notes and profiles 2. Agent layer - connect Hermes Agent to write files and run background scripts 3. Synthesis layer - use NotebookLM to create structured overviews from raw transcripts 4. Automation layer - trigger cron tasks to synchronize files and calendar standups The setup saves over $4,000 annually while keeping all private files offline Get the full step-by-step configuration guide and setup commands in the article below ↓

HE BUILT A 100% PRIVATE SECOND BRAIN IN OBSIDIAN TO AUTOMATE HIS RESEARCH USING HERMES AGENT AND NOTEBOOKLM He visualizes his entire knowledge graph on-premise without paying for cloud subscriptions By connecting Hermes Agent and NotebookLM, developers can index hundreds of documents and generate content locally Four components of this local knowledge architecture: 1. Memory layer - set up an Obsidian PARA vault to store markdown notes and profiles 2. Agent layer - connect Hermes Agent to write files and run background scripts 3. Synthesis layer - use NotebookLM to create structured overviews from raw transcripts 4. Automation layer - trigger cron tasks to synchronize files and calendar standups The setup saves over $4,000 annually while keeping all private files offline Get the full step-by-step configuration guide and setup commands in the article below ↓

marfin

20,684 Aufrufe • vor 1 Monat

day 23 of forking Paperclip today i added knowledge bases for projects and you company. based on Andrej Karpathy idea below. added a 'system chores' and 'agent chores' to manage memory. after each heartbeat run memories get added ('agent chores') and every 4 hours memeries get managed. This is in a db with an API for all agents to use and is seperated form the agents memeroes which they get at instantiation time.

day 23 of forking Paperclip today i added knowledge bases for projects and you company. based on Andrej Karpathy idea below. added a 'system chores' and 'agent chores' to manage memory. after each heartbeat run memories get added ('agent chores') and every 4 hours memeries get managed. This is in a db with an API for all agents to use and is seperated form the agents memeroes which they get at instantiation time.

Michael Houghton 🐺

18,188 Aufrufe • vor 3 Monaten

Simplicity is at the heart of great software. This is one of the reasons why Claude Code has been sticky for me. As a builder, I love planning and brainstorming, and this is now a key focus of Claude Code. I use Shift + Tab a lot to cycle between brainstorming, planning, and execution. This functionality provides the appropriate interface for me to either be very involved or less involved as I please. This works particularly well when building out new and complex features or entire new projects. This saves a huge amount of time. It allows me to tune Claude Code to execute and build more effectively. It also builds a loop of trust, and I often (surprisingly) find Claude Code asking for clarifications when it's confused. Coding agents don't normally do that. I have shared before on the power of brainstorming with AI for longer times. Try it and you will not be disappointed. Vibe coding is fun, but pair it with intentional development cycles, and you watch how far you can take a project with coding agents today.

Simplicity is at the heart of great software. This is one of the reasons why Claude Code has been sticky for me. As a builder, I love planning and brainstorming, and this is now a key focus of Claude Code. I use Shift + Tab a lot to cycle between brainstorming, planning, and execution. This functionality provides the appropriate interface for me to either be very involved or less involved as I please. This works particularly well when building out new and complex features or entire new projects. This saves a huge amount of time. It allows me to tune Claude Code to execute and build more effectively. It also builds a loop of trust, and I often (surprisingly) find Claude Code asking for clarifications when it's confused. Coding agents don't normally do that. I have shared before on the power of brainstorming with AI for longer times. Try it and you will not be disappointed. Vibe coding is fun, but pair it with intentional development cycles, and you watch how far you can take a project with coding agents today.

elvis

81,765 Aufrufe • vor 9 Monaten

Claude Design + Shopify is f*cking ridiculous 🤯 You can now publish pages from Claude Design → Claude Code → Shopify. Built 100% with Claude Design, Claude Code, and the Shopify CLI. Perfect for DTC brands and agencies who want to skip the design → dev handoff entirely. Here's how it works: → Design any landing page in Claude Design → Export as a zip and drop it into Claude Code → Install the Shopify + Shopify AI Toolkit plugins → Prompt Claude to convert the HTML into a Shopify page template + push to live theme → Claude uploads the images, deploys the files, and creates a published page No more handing designs off to a dev and waiting 2 weeks for a Shopify page. What you get: - A workflow that turns any Claude Design page into a real Shopify page template - Editable sections so your marketing team can swap copy, images, and CTAs without code - Images uploaded straight to Shopify Files automatically - A files-only deploy that only touches what's new in your live theme - A repeatable pipeline you can use every time you design a new landing page This is essentially the design-to-deploy pipeline brands have been waiting for. I put together a step-by-step playbook for going from Claude Design → published Shopify page. Every install, every plugin, every command, and the exact prompt that runs the whole thing. Want the playbook for free? > Like this post > Comment "SHOP" And I'll send it over (must be following so I can DM)

Claude Design + Shopify is f*cking ridiculous 🤯 You can now publish pages from Claude Design → Claude Code → Shopify. Built 100% with Claude Design, Claude Code, and the Shopify CLI. Perfect for DTC brands and agencies who want to skip the design → dev handoff entirely. Here's how it works: → Design any landing page in Claude Design → Export as a zip and drop it into Claude Code → Install the Shopify + Shopify AI Toolkit plugins → Prompt Claude to convert the HTML into a Shopify page template + push to live theme → Claude uploads the images, deploys the files, and creates a published page No more handing designs off to a dev and waiting 2 weeks for a Shopify page. What you get: - A workflow that turns any Claude Design page into a real Shopify page template - Editable sections so your marketing team can swap copy, images, and CTAs without code - Images uploaded straight to Shopify Files automatically - A files-only deploy that only touches what's new in your live theme - A repeatable pipeline you can use every time you design a new landing page This is essentially the design-to-deploy pipeline brands have been waiting for. I put together a step-by-step playbook for going from Claude Design → published Shopify page. Every install, every plugin, every command, and the exact prompt that runs the whole thing. Want the playbook for free? > Like this post > Comment "SHOP" And I'll send it over (must be following so I can DM)

Mike Futia

57,469 Aufrufe • vor 2 Monaten

“Hoverla, Ukraine. The highest point in Ukraine now looks like this. 2061 meters. When you stand here and hear the wind fluttering these flags, you feel both pride and incredible sadness. Behind each of them is someone's life, someone's pain and someone's feat. The guys leave them in memory of their comrades who will never see these mountains again. Someone climbs here on prostheses, overcoming themselves so that the flag of their unit flies closer to the sky. This is no longer just a tourist peak. This is a place of our strength and memory.”

“Hoverla, Ukraine. The highest point in Ukraine now looks like this. 2061 meters. When you stand here and hear the wind fluttering these flags, you feel both pride and incredible sadness. Behind each of them is someone's life, someone's pain and someone's feat. The guys leave them in memory of their comrades who will never see these mountains again. Someone climbs here on prostheses, overcoming themselves so that the flag of their unit flies closer to the sky. This is no longer just a tourist peak. This is a place of our strength and memory.”

Oksii ✚ 🇺🇦

198,792 Aufrufe • vor 3 Monaten

I genuinely don't understand why everyone isn't using this yet Andrej Karpathy, a co-founder of OpenAI, posted a simple idea that hit 16 million views: stop using AI to write code, use it to build a second brain. You point Claude Code at a folder, drop in any source, an article, a transcript, a PDF, and Claude reads it, links it, and files it into a living wiki of everything you know. It compounds like interest, the more you feed it, the smarter it gets. Here's the whole thing: > Install Obsidian, create a vault, open it in Claude Code > Paste Karpathy's wiki idea file and tell Claude to build it > Claude makes three folders: raw for sources, wiki for its pages, a CLAUDE.md that runs it > Drop any source into raw and say "ingest this" > Ask questions across everything, forever Five minutes to set up, and you never start from a blank chat again. Full step-by-step guide with Claude and Obsidian, link below. Bookmark this

I genuinely don't understand why everyone isn't using this yet Andrej Karpathy, a co-founder of OpenAI, posted a simple idea that hit 16 million views: stop using AI to write code, use it to build a second brain. You point Claude Code at a folder, drop in any source, an article, a transcript, a PDF, and Claude reads it, links it, and files it into a living wiki of everything you know. It compounds like interest, the more you feed it, the smarter it gets. Here's the whole thing: > Install Obsidian, create a vault, open it in Claude Code > Paste Karpathy's wiki idea file and tell Claude to build it > Claude makes three folders: raw for sources, wiki for its pages, a CLAUDE.md that runs it > Drop any source into raw and say "ingest this" > Ask questions across everything, forever Five minutes to set up, and you never start from a blank chat again. Full step-by-step guide with Claude and Obsidian, link below. Bookmark this

Ridark

6,917,862 Aufrufe • vor 1 Monat

this might actually be the new ugc method i've seen 100s of ai ugc ads lately where it's the same creator in every single one. same voice, same face, same brand colors, never drifts once what they're actually doing is running notch agents with memory they load the brand context one time. cloned voice, reference photo, brand color codes, do-not-say words, taste or they just teach it inline as they chat, and every correction becomes a saved rule then before it generates a single frame, the agent reads every relevant memory first. hard rules, reference assets, language constraints, taste preferences, all of it and it's workspace-wide, so every new ad already knows the brand before you type a word each chat isn't a re-introduction anymore. it's training your brand isn't a prompt. it's permanent context run 30-50 of those a month at a real offer and the volume does the rest punchier cut: the new method nobody's clocked yet run notch agents with memory on load the brand once. cloned voice, reference photo, color codes, do-not-say words, taste. or teach it inline and every correction saves as a rule it reads every relevant memory before it generates a single frame, workspace-wide each chat isn't a re-introduction. it's training. your brand isn't a prompt, it's permanent context run it at a real rate and the volume prints.

Sulfur

15,290 Aufrufe • vor 2 Monaten

President Trump just posted this on Truth Social honoring hockey legend Claude Lemieux: “Claude Lemieux, a true Legend of the Game, and one of the fiercest competitors Hockey has ever seen, has passed away. Claude was a friend to the family, and a tremendous ‘TRUMP’ supporter. He won 4 Stanley Cups with 3 different Teams — Montreal, New Jersey, and Colorado — His 80 Career Playoff Goals rank among the All Time Greats. Brendan, my thoughts are with you, Deborah, and the whole Lemieux family — You and your Dad were Warriors on the Ice. Claude will be missed by all who love Winning and Toughness.” Rest in Peace, Claude. A true warrior on and off the ice. Prayers up for Brendan, Deborah, and the entire Lemieux family. Claude was gone way too soon. He will always be remembered as a true legend. (Video: AI tribute to Claude Lemieux 🏒)

President Trump just posted this on Truth Social honoring hockey legend Claude Lemieux: “Claude Lemieux, a true Legend of the Game, and one of the fiercest competitors Hockey has ever seen, has passed away. Claude was a friend to the family, and a tremendous ‘TRUMP’ supporter. He won 4 Stanley Cups with 3 different Teams — Montreal, New Jersey, and Colorado — His 80 Career Playoff Goals rank among the All Time Greats. Brendan, my thoughts are with you, Deborah, and the whole Lemieux family — You and your Dad were Warriors on the Ice. Claude will be missed by all who love Winning and Toughness.” Rest in Peace, Claude. A true warrior on and off the ice. Prayers up for Brendan, Deborah, and the entire Lemieux family. Claude was gone way too soon. He will always be remembered as a true legend. (Video: AI tribute to Claude Lemieux 🏒)

Paul A. Szypula 🇺🇸

253,621 Aufrufe • vor 2 Monaten

Robots struggle with strict action rules…memory and symbols help them learn fast. [Project + Full video link ⬇️] Robots struggle when tasks require specific steps in a fixed order. What if memory helped them think symbolically and learn faster? Solving tasks like unlocking a door then opening it is hard for deep RL. But by learning constraint relationships and storing them in memory, robots can solve these tasks much faster; with fewer trials and less training. Why it works ✅ Learns symbolic rules about action constraints ✅ Uses memory to transfer what it learned across tasks ✅ Handles real-world exploration with just 30 minutes of data ✅ Needs 10x fewer episodes than deep RL approaches This memory-based method shows a promising path forward for robots learning structured, real-world tasks. Full video: Paper: Thank you, Mrinal Verghese for sharing this amazing work! 🙏

Robots struggle with strict action rules…memory and symbols help them learn fast. [Project + Full video link ⬇️] Robots struggle when tasks require specific steps in a fixed order. What if memory helped them think symbolically and learn faster? Solving tasks like unlocking a door then opening it is hard for deep RL. But by learning constraint relationships and storing them in memory, robots can solve these tasks much faster; with fewer trials and less training. Why it works ✅ Learns symbolic rules about action constraints ✅ Uses memory to transfer what it learned across tasks ✅ Handles real-world exploration with just 30 minutes of data ✅ Needs 10x fewer episodes than deep RL approaches This memory-based method shows a promising path forward for robots learning structured, real-world tasks. Full video: Paper: Thank you, Mrinal Verghese for sharing this amazing work! 🙏

Ilir Aliu - eu/acc

10,241 Aufrufe • vor 1 Jahr

$100,000/month net with fake girlfriends isn’t an offshore anymore, it’s a new SaaS wearing an OnlyFans skin this article shows how an 8-12 person OF agency becomes a solo op with four files per persona: persona.md, voice.md, flux.md, brain.md. the real asset is brain.md - memory on each fan that scales 5+ personas and prints $127,000 on ~$400 compute in the end, the "agency" is just a Claude code orchestrator with a cron job pinging five inboxes every 30 seconds

$100,000/month net with fake girlfriends isn’t an offshore anymore, it’s a new SaaS wearing an OnlyFans skin this article shows how an 8-12 person OF agency becomes a solo op with four files per persona: persona.md, voice.md, flux.md, brain.md. the real asset is brain.md - memory on each fan that scales 5+ personas and prints $127,000 on ~$400 compute in the end, the "agency" is just a Claude code orchestrator with a cron job pinging five inboxes every 30 seconds

Valentin

36,640 Aufrufe • vor 3 Monaten

ANTHROPIC JUST TURNED AI AGENTS INTO GIT REPOS Anthropic shipped "ant" - a CLI that runs every Claude API endpoint straight from your terminal. The headline isn't the terminal access. It's that you can now version-control an AI agent as YAML in Git and have CI sync it to the Claude Platform, the same way you ship code. - Every API resource is a subcommand: messages, models, files, agents, sessions - Define an agent in a YAML file, check it into your repo, and keep it in sync with one update command - Spin up a session, send it an event, then pull every event and tool call back from the same CLI - Claude Code knows how to drive ant out of the box - it shells out and reads the results with no glue code Agents just stopped being prompts you babysit and became infrastructure you deploy.

ANTHROPIC JUST TURNED AI AGENTS INTO GIT REPOS Anthropic shipped "ant" - a CLI that runs every Claude API endpoint straight from your terminal. The headline isn't the terminal access. It's that you can now version-control an AI agent as YAML in Git and have CI sync it to the Claude Platform, the same way you ship code. - Every API resource is a subcommand: messages, models, files, agents, sessions - Define an agent in a YAML file, check it into your repo, and keep it in sync with one update command - Spin up a session, send it an event, then pull every event and tool call back from the same CLI - Claude Code knows how to drive ant out of the box - it shells out and reads the results with no glue code Agents just stopped being prompts you babysit and became infrastructure you deploy.

BuBBliK

200,319 Aufrufe • vor 2 Monaten

Researchers made KMeans 200x faster. And the new technique also beats approaches like cuML and FAISS. Flash-KMeans is an IO-aware implementation of exact KMeans that redesigns the algorithm around modern GPU bottlenecks. By attacking the memory bottlenecks directly, Flash-KMeans achieves: - 33x speedup over cuML - 200x speedup over FAISS This speedup comes from how it moves through GPU memory. Standard KMeans runs in two steps, and both are bottlenecked by reads and writes to GPU memory: 1) The first step matches every point to its nearest centroid. Standard KMeans computes the full point-to-centroid distance matrix, writes it out to GPU memory, then reads it back to find each nearest centroid. That write-then-read round trip is the bottleneck. Flash-KMeans combines the distance calculation with the nearest-centroid step, so the result is computed on-chip and the full matrix is never written out. 2) The second step recomputes each centroid by averaging the points assigned to it. Standard KMeans has thousands of threads writing into the same centroid slots at once, so they stall waiting for their turn. Flash-KMeans sorts points by cluster first, turning scattered writes into sequential reductions that read and write memory in one efficient pass. Using these two optimizations at the million-scale, Flash-KMeans completes a standard KMeans iteration in a few milliseconds. The video below depicts this in action. Several reasons why this is important: KMeans has always been an offline primitive. Something you run once to preprocess data and move on. These speedups make the approach viable in several runtime-critical systems. ↳ Vector indices like FAISS use KMeans to build search indices. Faster KMeans means you can re-index dynamically as data changes. ↳ LLM quantization methods need KMeans to find optimal weight codebooks, per layer, repeatedly. What takes hours could now take minutes. ↳ MoE models need fast token routing at inference time. Flash-KMeans makes it viable to run this inside the inference loop, not just in preprocessing. I have shared the paper in the replies. That said, memory is the real constraint Flash-KMeans solves, and the problem is not just limited to clustering. The vectors a RAG system stores after indexing create similar bottlenecks. I wrote a detailed walkthrough recently on cutting this vector memory by 32x with binary quantization, querying 36M+ vectors in a few milliseconds. Read it below.

Researchers made KMeans 200x faster. And the new technique also beats approaches like cuML and FAISS. Flash-KMeans is an IO-aware implementation of exact KMeans that redesigns the algorithm around modern GPU bottlenecks. By attacking the memory bottlenecks directly, Flash-KMeans achieves: - 33x speedup over cuML - 200x speedup over FAISS This speedup comes from how it moves through GPU memory. Standard KMeans runs in two steps, and both are bottlenecked by reads and writes to GPU memory: 1) The first step matches every point to its nearest centroid. Standard KMeans computes the full point-to-centroid distance matrix, writes it out to GPU memory, then reads it back to find each nearest centroid. That write-then-read round trip is the bottleneck. Flash-KMeans combines the distance calculation with the nearest-centroid step, so the result is computed on-chip and the full matrix is never written out. 2) The second step recomputes each centroid by averaging the points assigned to it. Standard KMeans has thousands of threads writing into the same centroid slots at once, so they stall waiting for their turn. Flash-KMeans sorts points by cluster first, turning scattered writes into sequential reductions that read and write memory in one efficient pass. Using these two optimizations at the million-scale, Flash-KMeans completes a standard KMeans iteration in a few milliseconds. The video below depicts this in action. Several reasons why this is important: KMeans has always been an offline primitive. Something you run once to preprocess data and move on. These speedups make the approach viable in several runtime-critical systems. ↳ Vector indices like FAISS use KMeans to build search indices. Faster KMeans means you can re-index dynamically as data changes. ↳ LLM quantization methods need KMeans to find optimal weight codebooks, per layer, repeatedly. What takes hours could now take minutes. ↳ MoE models need fast token routing at inference time. Flash-KMeans makes it viable to run this inside the inference loop, not just in preprocessing. I have shared the paper in the replies. That said, memory is the real constraint Flash-KMeans solves, and the problem is not just limited to clustering. The vectors a RAG system stores after indexing create similar bottlenecks. I wrote a detailed walkthrough recently on cutting this vector memory by 32x with binary quantization, querying 36M+ vectors in a few milliseconds. Read it below.

Avi Chawla

89,234 Aufrufe • vor 1 Monat

An Asian guy discovered a method to learn any knowledge fast using AI. The key is Claude + Obsidian. Most people learn slowly: read, forget, read again, forget again. His method: use Claude to break any content into small, interconnected notes. Use Obsidian to link them, so every piece of knowledge is never isolated. Slow method: highlight books, keep going, forget after a week. Fast method: Claude breaks it into atomic notes, Obsidian links them into a network. Six months later, a new idea instantly connects to twenty things you already know. I compiled the full A–Z guide to building this second brain with Claude and Obsidian that most people have never discovered. Article below

An Asian guy discovered a method to learn any knowledge fast using AI. The key is Claude + Obsidian. Most people learn slowly: read, forget, read again, forget again. His method: use Claude to break any content into small, interconnected notes. Use Obsidian to link them, so every piece of knowledge is never isolated. Slow method: highlight books, keep going, forget after a week. Fast method: Claude breaks it into atomic notes, Obsidian links them into a network. Six months later, a new idea instantly connects to twenty things you already know. I compiled the full A–Z guide to building this second brain with Claude and Obsidian that most people have never discovered. Article below

Kirill

23,264 Aufrufe • vor 23 Tagen