@lovable @CUBoulder @cuatlas This is my submission for the Lovable hackathon featuring @AnthropicAI @supabase @elevenlabs and @eqtventures
English
Daniel Ritchie
3.5K posts
Daniel Ritchie
@deploydan
Yes Chef! | Brain Wave Collective
Denver, CO เข้าร่วม Ocak 2015
2.4K กำลังติดตาม617 ผู้ติดตาม
Check out the app I built with @lovable to showcase @CUBoulder student work from the @cuatlas GenAI class!
genai-showcase.lovable.app
English
@kwindla @arjunram @pipecat_ai We had fun navigating this. Although we couldn't get it across the line for the hackathon, I can definitely see some great use cases.
Ideal would be to plug in the framework and then just swap where the action happens. Can even imagine switching between local/remote services
English
I do think the next "big architecture" thinking we should do (collectively, as a community) for Pipecat is what the right abstractions are for local/cloud hybrid.
Pipecat 1.0 was all about getting the voice-to-voice latency, reliability, multi-modal frame types, interruption handling, phrase endpointing, flexible pipeline architecture, ability to use several models, ability to easily switch between models, etc to be really rock solid.
Pipecat 2.0 was about how you "drive" a real-time AI pipeline from a client with events/actions and LLM function calling. (See docs.rtvi.ai/introduction for an open spec for defining introspectible "services" and "actions" for real-time AI.)
Pipecat 3.0 probably should be about local/cloud hybrid use cases.
I was talking a little bit about that to @altryne, @JunaidDawud, and @deploydan during the hackathon last week. They were building a "local only" voice/vision pipeline for their robot. Even though we have a local transport type in Pipecat, it wasn't quite the right shape for the (really cool) thing they were building.
English
The text-to-speech and speech-to-text services in @pipecat_ai now support voice-to-voice LLM conversational AI in 79 languages.
People "outside the AI world" often ask me how much of the hype about AI is real.
One of the things I always say is that the global, multi-lingual, and translation capabilities of AI models are a big, big deal.
The Internet collapsed distance globally, but doesn't bridge language gaps and for a long time was very English-language-centric. I'm heartened, this time around, that a lot of people are doing a lot of great work training models in a lot of languages.
English
And it's Open Source!
github.com/brainwavecolle…
kwindla@kwindla
Second place: Foul-mouthed robot chef Abitious hackathon robot build and some seriously funny prompt engineering. SO-ARM 100 + Raspberry Pi + Porcupine wake word + @GoogleAI Gemini Vision + @cartesia TTS with a custom voice + @weights_biases Weave. By @altryne, @JunaidDawud, and @deploydan
English
@UnityEagle @JunaidDawud @altryne The gauntlet has been thrown!
I'm not going to say it doesn't matter whether or not we win, but we're definitely going to have the most fun
English
English
Heyo!!!! We took 2nd place among the remote teams!! Team “Yes Chef!!!” aka @deploydan @altryne and I are so pumped for this win!! Big thanks to all who organized, helped out, and judged.
Gahhhhh!!!!!!!
kwindla@kwindla
Winners from the Real-time Voice and Video Hackathon last weekend — remote teams division ...
English
Second Place?! Our trash talking chef may not be the standard use case for advanced robotic technology but the fact is we couldn't have done it without the SO-ARM100 by @therobotstudio and the @huggingface @LeRobotHF library by @RemiCadene
Let's Go!
kwindla@kwindla
Second place: Foul-mouthed robot chef Abitious hackathon robot build and some seriously funny prompt engineering. SO-ARM 100 + Raspberry Pi + Porcupine wake word + @GoogleAI Gemini Vision + @cartesia TTS with a custom voice + @weights_biases Weave. By @altryne, @JunaidDawud, and @deploydan
English
Team Yes Chef is happy to present: Robotic chef AI puppet that cooks you while cook 🧑🍳
(our submission to the Conversational Voice AI Hackathon 🤞)
We had such a blast making this one, from ideation to 3D printing, to experimental hardware, 👇 a breakdown and repro steps
kwindla@kwindla
Conversational Voice and Video AI Hackathon Oct 19th-20th - in-person in SF and remote $20,000 in cash prizes for the best ... 💠 voice AI agents 💠 virtual avatar experiences 💠 UIs for multi-modal AI 💠 apps built around conversational dynamics 💠 art projects 💠 anything else you dream up ... Sponsored by @cartesia @covaldev, @googlecloud, @tavus, @OracleCloud, @ProductHunt, @trydaily, and @Vapi_AI Sponsors are offering free credits and hands-on help with projects. Winners of the remote track get a pre-launch feedback session with the Product Hunt team. We will be hanging out all weekend at SolarisAI in San Francisco and on Discord. Join us from anywhere in the world. Start a new project, or bring a project you've been working on and add a new real-time feature. All projects are eligible for prizes: local, remote, new, old. Sign up here: lu.ma/6www8b0t
English
@altryne @LeRobotHF @RemiCadene Also shout out to @therobotstudio for a very cool open source design of this incredible #airobot arm. #realcoolrobots #the_robot_studio #robot
English
Assembling the pieces took a while, but so does figuring out how to actually control whatever we wanted to do.
See the @LeRobotHF SO-ARM100 is still experimental, huge shoutout to @RemiCadene who was updating the library as we were building!
English
We used @wandb Weave to track not only the LLM text, but also the images and the audio, as Weave is one of the first observability platforms that supports those modalities in the same table!
Here's our Weave dashboard with some sick 🔞 burns
wandb.ai/wandb/yes-chef…
English
English
English
@wizardai_m @altryne We both had @LeRobotHF @therobotstudio arms for next weekend, and @altryne's chef puppet for an unrelated project.
Thought it would be fun to combing and hilarious if the robot insulted you. Now everyone can experience the joys of having @GordonRamsay in their kitchen.
English
Daniel Ritchie รีทวีตแล้ว
I don't want to say "I told you so", but I told you so.
Repeatedly.
I was heavily criticized for it. Mocked, even.
François Chollet@fchollet
People have been rewriting history and saying that "everyone has always believed that LLMs alone wouldn't be AGI and that extensive scaffolding around them would be necessary". No, throughout most of 2023 (the "sparks of AGI" era) the mainstream bay area belief was that LLMs were *already* AGI, and that merely scaling their parameter count and training data size by ~2 OOM without changing anything else would lead to super-intelligence.
English
Daniel Ritchie รีทวีตแล้ว
Omg I haven't belly laughed like this in a while 🤩 😂
Yall are not ready for this, we haven't even finished and this is one of the craziest things I've even gotten to participate in 🧑🍳
@deploydan @JunaidDawud
Alex Volkov@altryne
Let's gooo - we hacking something supercool remotely for this with @JunaidDawud and @deploydan AKA team Yes CHEF h/t @LeRobotHF
English
Daniel Ritchie รีทวีตแล้ว
Let's gooo - we hacking something supercool remotely for this with @JunaidDawud and @deploydan
AKA team Yes CHEF
h/t @LeRobotHF
kwindla@kwindla
We just passed 5⃣0⃣0⃣ sign-ups for the Open Source Real-time Voice and Video AI Hackathon this weekend. It's not too late to sign-up and join the fun — in SF and remote on Discord. Now is a great time to thank our volunteer judges for joining us on Sunday to check out all the projects (and to award the $20,000 in *cash* prizes.) @jheitzeb — global organizer of AI Tinkerers, founder, ai wrangler @aerbitsai — bringing ai video to life at Tavus @lizziepika — dev advocate and ai accelerator at Cloudflare @charles_irl — ai engineer extraordinaire at Modal @khushkhushkhush — investor at A* Adarsh Sivadas — Google & Cloud at Google @nehiljain — founder of DemoDrive AI @hackgoofer — stealth startup founder (and winner of 7 hackathons in the last ~year, so it's lucky for the rest of us she's a judge this time)
English
Come hack August 2nd-4th
@GroqInc @CodeTechTalent @AITinkerers @twelve_labs
#Denver #Boulder #AI #hackathon #siggraph
English
