dbish

18 ms·

211.
by dbish 3y ago
Kids are funny. I’ve seen multiple go through being completely obsessed up to about 5 to losing interest and never looking at a model train again. I think there’s something about that young age that makes them particularly interested, but t
212.
by dbish 3y ago
Fwiw, we've found the Azure AI OCR service to be pretty good, much better then anything we could get from Tesseract out of the box (no tuning).
213.
by dbish 3y ago
I definitely think kids need fathers around (I am a father myself and still debating the level of involvement that is useful or not). I’m just talking about the somewhat unavailable father of the 50s vs. parents of today who flip flop betwe
214.
by dbish 3y ago
There’s no reason to use numbers then. Use words and don’t hide behind sounding precise and mathematical
215.
by dbish 3y ago
It’s also barely clarified and has a ton of false precision problems, which I’d expect a researcher who works in the field not to have. Why 10% vs 20% or 15%, no real reason, just vibes, but it gets clicks.
216.
by dbish 3y ago
We’re working on a tool that “screen shares” with AI and allows both looking at what is on screen when you’re sharing and acting on it (simulate clicks and other). What’s your use case? Happy to share more about where we’re at with the proj
217.
by dbish 3y ago
Better in the sense that you didn’t learn to just passively consume from a screen and learned to be independent. I’m not convinced that parents becoming involved to the extent they are is actually a net positive. Kids don’t learn to be on t
218.
by dbish 3y ago
The teaching part is what matters, it’s training (tuning in this case) a model, not just using a model already trained for inference (which is what I assume you mean). You’re providing new data that is used to update the model. Inference ac
219.
by dbish 3y ago
Looks cool. I’ll try it out when I’m on a Mac, mostly use Windows for daily dev right now. Been working on a similar side project based on something I prototyped when I used to work on Alexa but for windows first then likely going to be mul
220.
by dbish 3y ago
Still used in places that require a specific type of data privacy (ex:personal ads recommendations or keyboard predictions from usage), but it has a lot of overhead, is slower then other non federated processes, and can’t keep up with a cen
221.
by dbish 3y ago
Unknown words can’t really be handled other than mapping to a placeholder “unknown” vector in most cases. This can be dealt with in a variety of ways, text preprocessing being one of them
222.
by dbish 3y ago
Just use your embeddings model of choice that works with your language, I believe Ada from openai is multilingual but I don’t know what languages it works well on, there are many embedding models out there, huggingface is your friend in thi
223.
by dbish 3y ago
Maybe, seems like overkill to solve a relatively straight forward problemt that is about resolution to real entities with a generative model that has no entity grounding (and may introduce other problems). It's really just not solved b
224.
by dbish 3y ago
Many big “open source” releases in the AI community recently are not licensed for commercial use. Not really OSS at that point (ex:fuyu model from adept)
225.
by dbish 3y ago
This is part of the mismatch between comparing Alexa/Siri/Cortana to a chat based LLM right now. If you just want chat and info retrieval, today’s LLMs are way better then the conversational dialogue, search, and q&a capabilit
226.
by dbish 3y ago
Nice. Been trying to get it to make more classic chiptunes style sounds but can’t quite get there
227.
by dbish 3y ago
Also hard to say it’s really true. OpenAI is certainly, is Microsoft without OpenAI’s tech eating Google’s lunch?
228.
by dbish 3y ago
Also what does "a lot more lightweight and faster to use" mean when contrasting against Loom for you? Is their client particularly heavyweight in any way?
229.
by dbish 3y ago
Looks interesting conceptually to tackle an OSS option here with Atlassian swooping in on Loom, though there isn't much code yet, been working on a loom alternative as well, and debating open sourcing it. What do you have working so fa
230.
by dbish 3y ago
Would love to hear more about what you're trying to build here. Are you thinking use VNC to control and stream a desktop?
231.
by dbish 3y ago
This is like saying "I can't watch the video without a browser that supports video playback", yep?
232.
by dbish 3y ago
Pretty cool. I wonder if there's enough interest from other devs to hook into something like this or build "apps" on top of this ability. Are you trying to "productionize" or build a platform for others on this or w
233.
by dbish 3y ago
We know how the glasshole situation went :) Siri question type things I can use my watch for real time translation is neat but it has to be a big and frequent use case to differentiate from just taking my phone out and using google translat
234.
by dbish 3y ago
Seems like a design choice but it was odd. Both people didn't smile and seemed bored as you say. A bit odd, maybe tied to their like of things like Fashion Week where it's more of a serious look then a happy one
235.
by dbish 3y ago
Apple watch? Cellular mode allows this, has siri built in, can handle calling/messaging/etc. People don't want to replace their phones though.
236.
by dbish 3y ago
Not to mention that voice based Q&A has been done by everyone at this point. Should have focused on the differentiators they have with that form factor.
237.
by dbish 3y ago
Also more expensive, i pay 10/month for a dedicated watch, and i can still make 3rd part apps for it, i can't do that with humane as far as i can tell and don't really want to put it on my shirt like this. Only real different
238.
by dbish 3y ago
I don’t think people actually want this even though they talk about it unless it’s basically perfect which the tech still isn’t there for. It’s like the “book a flight” use case that has been used or eons, unless you’re rich (in which case
239.
by dbish 3y ago
Matters a lot for any company trying to use them. There are licensing agreements, security audits, legal approval, etc. that make using any of these proprietary license models quite hard if not impossible
240.
by dbish 3y ago
This is basically every virtual assistant’s pitch since forever, including Cortana, Siri, and Alexa. They’ve all tried some version of this, with almost no customer usage (most famous 3rd kart friendly one being Alexa skills).

More ›