What if every woman / person age 18 - 29 voted in #PA / #MI / #WI? I just made this animated simulator to motivate friends to vote https://t.co/0iWTtx3mjO
@bchydro Another planned power outage today in V7R? This is unacceptable - the second one in about 1 month. This is not the norm elsewhere in Canada/US. Many of us work from home and have young children who need heat. Please compensate.
@0xpff @BillHeyman@danshipper Can confirm from my testing it does pay attention, particularly if you use it to provide external data as context. I think the key word there is “strong” attention — the user and assistant messages are the strongest drivers of what the model will say next
@tunguz These 3 startups seem to be aiming for exactly that:
https://t.co/wxoz9PCsvA
https://t.co/yq6pAEdMPa
https://t.co/wbaYbW83Ec
I've built an internal version of similar which I'm trialing with employees at my work, but nice that there's turnkey options
@akshayerathi@nikhilkamathcio Agreed google meet is best UX, but have you managed to screenshare while getting an overlay of the person you’re talking to? Seems like a missing essential feature
@frantzfries Love this - I built a prototype of this a couple ELD ago. Testing it internally with sales team employees at the company I work for. DM me if you want to chat about this
@stef Definitely interested. I’ve built a barebones version of this for training my firm’s employees on domain specific material (in my case, our SaaS product for industrial fatigue mgmt). DM me if you want to chat or see a demo!
@danshipper Hi Dan - big fan, happy to chat. I’ve built a GPT chatbot for my work (domain specific answers on SaaS product) following the model for Huberman bot. Been really cool to use ChatGPT to help me translate Python docs to Node.js where I’m a slightly more comfortable newbie.
@OfficialLoganK@OpenAI I’d be happy to. I’ve recently made a chatbot that I’m piloting with employees at my company using GPT3 embeddings to serve up answers to domain-specific questions. It’s a fascinating moment in time!
@prundin@amanjha__ I’ve been wondering the exact same thing.
Will dig into whether fine tunings vs. embeddings is a better fit for letting GPT answer questions based on a corpus of specific concepts.
Seems like this is Embeddings territory but not 100% sure just yet. @sama ?
@AirCanada No, this was at general baggage claim so there was no Air Canada agent present, nor any other relevant staff. Can you provide us a courtesy voucher of some kind?