Pinned
Dee
2,931 posts
Our setup inside the WorkPod. Which one do you prefer? ๐ Mann, look at how amazing this WorkPod is!
- Replying to @work_loudertop-notch! ๐ I'd love to hear your thoughts on my product ideas too
- I ran gpt-oss-20b on dual rtx 5090s โก๏ธ with llama.cpp, hit ~3719 tok/s on prompt tests (pp512 @ d128)๐ Even at d1024, pp holds strong (~3490 tok/s). TG tests (128 gen tokens) slower at ~240โ261 tok/s. Solid speeds across configs! See the how-to below.๐งต
- Hype after Hype, guys. I seriously need your help to name this desk ๐ฅ๐ฅ Quick brief: 1-leg standing desk, integrated power outlet & wireless charging, touch control panel, high quality tempered glass, wipeable, move around easily. The chosen winner with the best name will
- Replying to @ns123abcSo, Sam Altman thinks AGI is a solved problem and we're already on to superintelligence? Sounds like someone's been hitting the AI power-ups a little too hard.
- โThe first rule is not to lose. The second rule is not to forget the first rule.โ โ Warren Buffett #WednesdayMotivation I'm trying my best and not giving up. Can I ask for your follow to our new account @autonomous_ny ๐ Thanks a lot and truly appreciate it!
- im done with cloudDone with cloud costs? ๐ฅ Brainyโs got human-like smarts and raw RTX 4090s power for your AI projects locally!
- Qwen3-30B running at ~3929 tok/s on EdgeAI Computer (2 x 4090D) with llama.cpp โก๏ธ
00:00 - Our RTX 4090 supercomputer is free to test for a week! Each go is a 12-hour slot. You can see how your AI runs on some proper hardware. Peep the schedule down there if you wanna book some time on Brainy! ๐
- Life of a Product Manager... Watching my 'child' grow ๐ And so proud to have my dream team here ๐
- Our bamboo top is so limited... Coz you can tell from its quality ๐Finally unboxed and fixed my desk from @autonomousdotai ๐คฉ ๐ Use 22DAVID2 to get $20 off on your orders from $499
- Context test: gpt-oss-20b ร @ollama on EdgeAI Computer (2ร RTX 4090D) Saw your post, had to run it @ivanfioravanti. Props to @ollama โ smooth experience โก๏ธ Results ๐
Ollama (amazing job!) context test of gpt-oss:20b ๐ฅ On Mac Studio 512GB here. Perfect implementation of the MXFP4 in collaboration with OpenAI! "Ollama is supporting the MXFP4 format natively without additional quantizations or conversions. New kernels are developed for - First time hosting this GiveAway for my followers ๐ฅ 1. FL @autonomousdotai & me ๐ 2. Like + RT 3. Reply 1 innovative product you want me to make Ends on 21 Dec. For US only Note: I will deliver the ErgoChair Curve to your house by Jan 2024!
