swein

@9mmsmg Clearly you've never heard of brown belts and tacos

Drove by a Taco Bell, and there was actually a line at the drive-through. How are people eating this stuff? Fast food is trash in general, but Taco Bell is bottom rung for me. Wouldn't eat it if it were free. What's wrong with you people?

@WallStreetMav India.

If X is not earning any ad revenue from Pakistan or Nigeria ... maybe users in those countries shouldn't be monetized? Just an idea.

Hopefully something better related to the algo is coming. We all love X, because it is better than the other options especially for free speech.

@thdxr @reem_a 37signals hey?

@reem_a hey

I'm hiring someone to join my team at Anthropic to lead Claude Code comms. This is not a role for someone who wants to run an old playbook. You'll need to be a Claude Code super user, understand developers and dev tools, and have great taste. You'll work hard, learn a lot, and ship with the best people around. Non-traditional comms paths welcome. My DMs are open!

@LLMJunky @TheAhmadOsman @0xSero @ivanfioravanti @ZenMagnets @NVIDIAAI Id send you private link to 6000 owners discord but cant seem to dm. Youll love the cards

Finally proud to announce that I've joined the GPU Minor Leagues. 2 x RTX 6000 Pro. I have six months to pay off the second GPU lol. You are all TERRIBLE influences.

@LundukeJournal Written in rust. They/them, blah blah, xyz is anti-semetic. It's all so tiresome, fights against one cause while pushing his own antisemetism claims left and right.

@LundukeJournal Lunduke i agree with you often but are you Jewish? You keep saying anti semetic often just trying to understand why?

Who is in charge of the releases of the GNOME Desktop Environment? According to the GNOME Release Calendar, two of the most prominent release managers (responsible for some of the most significant releases) are: Jeremy Bicha (jbicha) - A registered sex offender who, according to reports, was responsible for “thousands” of assaults of small children. Jordan Petridis (alatiera) - A man who attacks Jews, on the official GNOME website, as “apartheid, ethnosupremacist, babykillers” and “fascist maggots”. Both men have also defaced the source control repositories and wiki pages of “competing” open source projects, calling those they disagree with “Nazis”. At no point has GNOME made a statement against these actions, nor has GNOME acted to protect their contributors and users from the predators running their project. This is GNOME. release.gnome.org/calendar/

@Grummz Just needs some dlss 5

I don't think the new Lara Croft is working out so well.

@AMD @__tinygrad__ you got this?

Join the GPU MODE Hackathon, sponsored by AMD, and push the boundaries of LLM inference performance on leading open models, optimized for AMD Instinct MI355X GPUs. Finalists will compete for the $1.1M total cash prize pool across two independent tracks, each focused on a specific model and inference stack. Learn more and get registered here: luma.com/cqq4mojz

@DocManhattan777 @ArchetypeTheory Do it for Indians who have come here due to horrible Visa laws. They take white collar jobs while having lower IQ's and fraudulent experience/degrees. They hire based on cultural/racial nepotism. Sound familiar?

@ArchetypeTheory Now do Jewish representation to US GDP and Taxation revenue😉

15,000 Jews are in the American military 23,000 American Jews serve in the Israeli IDF. They are 4x under represented in the American military compared to other demographics Yet Jews are 35x over represented in government beauracracy with 70 percent of government cabinet positions going to Jews. America is occupied by a global terrorist mafia.

@TheAhmadOsman @mweinbach Building off ahmad’s comment. Vllm and sglang have superior prefill speed and handle kv cache much better. Say you want 4 concurrent or 32 sessions llamacpp quarters or context / 32 per session. Llamacpp is for cpu offload

@mweinbach Inference Engines have different optimizations for different hardware Most optimized for DGX Spark: TensorRT-LLM Inference Engines MATTER and they are NOT EQUAL (e.g. blogpost below) Opensource models ARE NOT just a memory size issue BTW ahmadosman.com/blog/do-not-us…

DGX Spark uses unified memory > 273 GB/s RTX PRO 6000 delivers > 1.8 TB/s (1792 GB/s) If someone told you they’re comparable, they’re wrong And this is exactly why llama.cpp isn’t the right tool here Try vLLM or SGLang on a GPU and you’ll see very different results

@akshay_elavia @michaelfreedman Even in India this would be a huge step up for file system ops and transaction logs

@michaelfreedman do you think this will help if i have a bunch of code repos cloned locally, and i want my agents to navigate and work across them? if not, any pointers for my use case?

Introducing TigerFS - a filesystem backed by PostgreSQL, and a filesystem interface to PostgreSQL. Idea is simple: Agents don't need fancy APIs or SDKs, they love the file system. ls, cat, find, grep. Pipelined UNIX tools. So let’s make files transactional and concurrent by backing them with a real database. There are two ways to use it: File-first: Write markdown, organize into directories. Writes are atomic, everything is auto-versioned. Any tool that works with files -- Claude Code, Cursor, grep, emacs -- just works. Multi-agent task coordination is just mv'ing files between todo/doing/done directories. Data-first: Mount any Postgres database and explore it with Unix tools. For large databases, chain filters into paths that push down to SQL: .by/customer_id/123/.order/created_at/.last/10/.export/json. Bulk import/export, no SQL needed, and ships with Claude Code skills. Every file is a real PostgreSQL row. Multiple agents and humans read and write concurrently with full ACID guarantees. The filesystem /is/ the API. Mounts via FUSE on Linux and NFS on macOS, no extra dependencies. Point it at an existing Postgres database, or spin up a free one on Tiger Cloud or Ghost. I built this mostly for agent workflows, but curious what else people would use it for. It's early but the core is solid. Feedback welcome. tigerfs.io

@DiggingInTheDi1 @TheBabylonBee Babylon bee is ran by jews. /shrug. defend their own I guess

@TheBabylonBee Hey, how many tours in the Middle East have the 100 lb soyboy Babylon Bee staff collectively done?

Man Bravely Resigns From Trump Administration At Great Cost Of Doing Podcast Tour, Getting Book Deal buff.ly/0mmN6EU

@ahmetb Based on the ADL 99.99% of everyone is an antisemite . Lol

i always knew python was anti-semantic

@swampratwatch @InsiderWire Banning africa, israel, and “south asia” is really all thats needed to have a real global voice without the spammers

@InsiderWire The convenience you demanded will be mandatory soon. Governments want compartmentalized internet so they can more freely propagandize their citizens, keep scandals localized, and shield themselves from criticism coming from people outside their jurisdiction.

#BREAKING: 𝕏 will soon let users restrict both posts and replies by region or country.

@0xSero Did the same prompt but night time with fire flies . qwen3.5 27B fp8

Holy moly, MiniMax-M2.7 is amazing, watch till the end.

@TheAhmadOsman @CryptoElite007 Agreed. What do you want a dgx spark +? More ram

@CryptoElite007 Different things, very different Unified Memory is NOT a competitor at all to NVIDIA’s offerings

Jensen just answered my question and ended it with saying: I am committed to open models Video as soon as the Q&A ends

@LottoLabs Already ahead of ya. 70tps on FP8 batch 1. Mtp3 vllm same model and card

Hermes Agent and qwen 3.5 27b on RTX6000 getting a nice 48TPS on vLLM Testing 3090 + vLLM next

@ARomeoSierra @MatthewBerman @nvidia @Dell AWQ or nvfp4 will sit just fine.

@MatthewBerman @nvidia @Dell Qwen 3.5 379B. Oh nm you don’t have enough RAM.

.@nvidia hand delivered a pre-production unit of the @Dell Pro Max with GB300 to my house. 100lbs beast with 750GB+ of unified memory to power the best open-source models in the world. What should I test first?

@levelsio it's only a few hotkey to remember, easier than vim

I hate tmux It's so incredibly user unfriendly The shortcuts make no sense I wish someone would make a better tmux Even just logging into tmux attaching the screen is an illogical hell to type Again I hate tmux, it's so shit

@gemsnper @0xSero 96GB VRAm, and a 4 bit quant in VLLM will run multiple sessions fast I'm sure. Just like qwen3.5 122B

@0xSero do you think it'll handle multi-agent systems well in a local setup? curious about its scalability on non-dedicated hardware.

I think we finally got a banger model from Mistral we can run locally FAST. This is sooooo exciting. They built it for complex math, agentic, and coding. huggingface.co/mistralai/Lean… I will have quants and reaps up for this by end of week.