@PezRadar Happy to see Blizzard reacting to bugs pretty fast! I was a bit afraid the lunar event would be full of infinitely scaling damage and players executing themselves ^^
Thanks everyone for liking and sharing the #Alpaca OPi5 demo 😃. I'll use this opportunity to address @MetaAI who developed and released the base model:
What if you'd change the license of #LLaMA 7B + 13B to CC BY 4.0 and become legends of really #open#offline#private#AI? 😎
Next round 😁🤖💬: #Alpaca (Stanford's fine tuned #LLaMA 7B) running on Orange Pi 5 8GB (#ARM RK3588S SBC) via llama.cpp. Real time video from load to finish (no cherry-picking) - Albert Einstein and Yoda's thoughts 😄 - @ggerganov@RichardSocher#reallyopen#ChatGPT
Meta's #LLaMA 7B 4-bit model running on #OrangePi 5 8GB (#ARM RK3588S SoC) via llama.cpp by @ggerganov. Real-time video including model loading!! 😲🤩🚀 (~7 Tokens/s)
Huge Roundup This week! 🔥
We cooked a complete recap of last week Capcom Cup IX !
👇Dive in for Capcom Cup action!👇
📺https://t.co/J0snTjnvP4
Watch and hit subscribe, we drop that good stuff REGULARLY
#CCIX#CPT2022
#ChatGPT for #Robotics - Microsoft implements a high-level API to control robots, then let's ChatGPT write code based on a task description like "warm up my lunch" 🙂🤖 - no magic, basically promt engineering 😉: https://t.co/xDrJJ8JZ4E - Video: https://t.co/wrO1iItvKB
I've tested various ports of @OpenAI#Whisper speech recognition to see how fast it can be on #ARM SBCs like #RaspberryPi 4 and #OrangePi 5: https://t.co/zvRuni631M - Spoiler: it can be very fast 😁🚀
SEPIA #OpenAssistant has been featured in the German #RaspberryPiGeek magazine 12/2022 together with @mycroft_ai 🤓✌️🤖 . The article talks about cloud-free, customizable, digital assistants as alternatives to Alexa: https://t.co/jOhAbbC8gy
SEPIA #OpenAssistant v2.7.0 has been released earlier this month ✌️🤖 and comes with a theme editor to build custom skins 🖌️🎨 long awaited support for #homeassistant 🗣️🏠💡 the ability to swap #speechrecognition models on-the-fly and much more! Check out: https://t.co/n5SCABm2Xr
Just something I wanted to share because I still love #Java 😃. Just wish it was more popular for #ML 😅.
"Popularity doesn’t necessarily translate to demand by employers. [...] we explored the sum of monthly active tests mandating a specific language." - @hackerrank
Even after 13 major releases #Android as OS still feels like a beta version 😑, every year: broken APIs, changed policies, different permissions, etc. Don't get me wrong, I love the possibilities and I'm usually happy with the result but keeping an app alive is exhausting 😵💫
Hey @Apple#Safari or @webkit teams, will we ever get solid microphone support for #PWA ? 😏 In Safari I get asked every few seconds and in a PWA I get asked once (on every! start) and then it breaks after a while ☹️. #Chromium based browsers are so far ahead here #WebApp#WebDev
Ok Whisper by @OpenAI is really interesting. I've tried the "tiny" (72MB) and "base" (139MB) models on #raspberrypi !! "tiny" takes 20s and "base" 47s for 17s recording. Both models work almost perfectly on noisy German AND English audio 😲😎. We need a streaming model next 🤤!
Experimenting with @home_assistant support for SEPIA #OpenAssistant 🗣🤖💡using cloud-free speech recognition server on @Raspberry_Pi connected to SEPIA web-client (PWA on the right with Proto skin). Left: HA web UI showing my #smarthome devices in real-time 🙂
The SEPIA offline #speechrecognition server v1.0.0 has been released 🥳 and fully supports @coqui_ai ASR models now 🐸🙂. It also comes with the latest Vosk version by @alphacep and a new Python client 🐍. 2 engines, hot-swap, same API, web and local 🤖😎 https://t.co/N5OGO7EF5s