@carlosthedwarf_ @warriorsworld 6'11 and 255, maybe not big enough for C and may move to PF later on in career.
English
Michael Quinlan
860 posts
Michael Quinlan
@michaeljquinlan
Machine learning, robotics and basketball. — Formerly: Everyday Robots (https://t.co/fIkQmAhRcb) / Google X, Clover
CA Katılım Nisan 2009
257 Takip Edilen181 Takipçiler
@ChongZitaZhang Yep, curating and labelling your own dataset is very eye opening. From making 50/50 decisions and realizing how hard it is to define or how many errors already exist, through to the monotony of it. Everyone doing ML should do it at least once
English
@krishnanrohit Lots of Australian highschools had these for gyms .. architectureau.com/articles/Binis…
English
Genuine question, could you just make a version of the dome that is the building?
Indian Tech & Infra@IndianTechGuide
🚨 China has built a 50m-tall inflatable dome over a construction site in Jinan to protect the surroundings from dust and noise.
English
@simonw @eckely You might enjoy this 2022 Wittgenstein ai app, wittgenstein.app/W39p1FHtmmaP83… , both impressive it's from 2022 and also have far we have come since
English
Confession: despite all of the debates about whether or not an LLM can "reason", I still don't really understand exactly what the term "reasoning" means
So just like with "agents" and "AI" itself, I'm not sure the people engaged in those debates are talking about the same thing
English
@OfficialLoganK Hey Logan, we tried these and they look good. Any ideas on if/when these will move out of preview? And do you know if there are plans to support more regions then central-1 for previews?
English
Today, we are rolling out three experimental models:
- A new smaller variant, Gemini 1.5 Flash-8B
- A stronger Gemini 1.5 Pro model (better on coding & complex prompts)
- A significantly improved Gemini 1.5 Flash model
Try them on aistudio.google.com, details in 🧵
English
@OfficialLoganK @levelsio It's ironic that vertex does have an endpoint you can point the openai sdk at.
There shouldn't be two packages with different call patterns. We need one Google wide sdk that handles both backends.
The AI studio team should lead on this and just make the compatibility layer.
English
@levelsio Will get OAI compatibility added to the SDK, AI Studio and Gemini API are for developers, Vertex is for Enterprise (like Azure).
English
Google APIs are the worst APIs on the web to work with any dev will agree
What's worse is that now we see that is actually affecting their bottom line: nobody is using their AI APIs cause they're too cumbersome to set up
Open AI's API just works with one Curl request
Brendan Dolan-Gavitt@moyix
OpenAI: pip install openai and set OPENAI_API_KEY Anthropic: yea same but s/openai/anthropic/g Google: oh boy. ok so you have a GCP account? no? ok go set that up. and a payment method. now make a "project". SURVEY POPUP! k now gcloud auth. wait you have the gcloud CLI right–
English
@altryne @aidan_mclau @moyix @googleaistudio Unfortunately GCP credits only work with vertex calls, even if you bill the aistudio key through your GCP project id.
Or at least the case for me. Mind boggles, I guess vertex belongs to cloud, AI studio to another org so they won't share credits or something.
English
OpenAI: pip install openai and set OPENAI_API_KEY
Anthropic: yea same but s/openai/anthropic/g
Google: oh boy. ok so you have a GCP account? no? ok go set that up. and a payment method. now make a "project". SURVEY POPUP! k now gcloud auth. wait you have the gcloud CLI right–
English
@carlosthedwarf_ @Julxxx0 Season 1 still holds up .. I rewatched some of it again a few years ago
English
@simonw I'm hoping they can keep their eval set fresh (or hidden enough) that it reduces the chances of getting p-hacked.
English
Under-appreciated leaderboard in my opinion - function calling (aka tool usage) is one of the most interesting applications of LLMs, plus it should be independent of how much "knowledge" is baked into the models
Shishir Patil@shishirpatil_
📊Delighted to welcome Command-R-Plus, Llama-3, and and Gemini-Pro-1.5 into the Berkeley Function Calling Leaderboard. Check out how they stack up across different categories, P95 latency, and costs at gorilla.cs.berkeley.edu/leaderboard.ht… Congratulations to @cohere, @AIatMeta, and @GoogleDeepMind for pushing the Function Calling/Tool Calling boundaries even further 🚀
English
@john_d_beatty @sampullara The under 18 part is very odd.
English
@sampullara The nanny rails are strong with this one... I'm getting an incredible number of refusals now on all sorts of mundane tasks. Need to try the API.
English
C++ considered harmful
fox@soft_fox_lad
Gemini is apparently told your Google account age and will answer questions w/the appropriate caution. Here it is refusing to help someone with C++ because they’re he’s under 18, and advanced C++ is a danger to a young mind. gemini.google.com/share/23803238… From news.ycombinator.com/item?id=393834…
English
Super happy for Caitlin Clark! Couldn't think of a better person to break the record
Michael Quinlan@michaeljquinlan
Here is my annual Pistol Pete tweet ... Amazing that these highlights are almost 50 years old. youtu.be/n8qUZILi8IM
English
@carlosthedwarf_ Yeah definitely obstruction (if that's the change you were referring too). The second player moved across the field solely to stand next to someone and make a double wide block.
English
Very happy with that rule change. Of course Ennis wants to leave it as is🤡
English
@marioprats Although i 100% understand the fun factor !
English
@marioprats Gemma is worse than Gemini or Chatgpt ... I'd start there before jumping to Gemma
English
I've been playing with Gemma today. Interestingly having it working in C++ was way easier than python / colab! github.com/google/gemma.c…
English
@ntarun You should try and checkout Apptronik if you haven't already
English
It’s crazy that someone would even sell us one of these humanoid robots to us consumers because RaaS aka Robot Slavery seems like the better business model.
Tarun@ntarun
What comes after AI? Robotics. Robotics (led by Humanoid Robotics) likely will turn out to be the largest category of the US and World economy, almost by definition as it can replace all forms of human labor and will be the cheapest way to move physical atoms.
Austin, TX 🇺🇸 English
@__daave Pro Git is in the correct location, below comfortable reaching height
English
I’ll be posting more updates as I figure things out. First thing is turning this space into a lab.
English
Not to steal @karpathy’s spotlight (more like me tripping on a wandering photon from the sun, in this case), but I was going to post that same post.
Friday is my official last day.
I’m leaving to focus on personal projects that I managed to fund 🙏
Andrej Karpathy@karpathy
Hi everyone yes, I left OpenAI yesterday. First of all nothing "happened" and it’s not a result of any particular event, issue or drama (but please keep the conspiracy theories coming as they are highly entertaining :)). Actually, being at OpenAI over the last ~year has been really great - the team is really strong, the people are wonderful, and the roadmap is very exciting, and I think we all have a lot to look forward to. My immediate plan is to work on my personal projects and see what happens. Those of you who’ve followed me for a while may have a sense for what that might look like ;) Cheers
English
@krishnanrohit I also had reasonable luck with tortoise-tts last October-ish (i.e before the openai API). It was fairly good. I only used it on Linux with GPU, can't speak for performance on Mac or a more limited system.
English
@tstorm81 Haha thought about it! But figured with the car being only C++ it pushed the lua just low enough.
English
88-92: Basic
92-96: Turbo Pascal
97-98: C++
99-01: Java
02-03: C++
04-07: C++ & Python
08-10: C++
11: Objective C
12-15: Java
16-19: C++
20-22: C++ & Python
23: Python
Steve Krouse@stevekrouse
My main programming language: 2006 logo 2007 logo 2008 scheme 2009 haskell 2010 java 2011 haskell 2012 haskell 2013 python 2014 ruby 2015 scratch 2016 javascript 2017 javascript 2018 haskell 2019 haskell 2020 ruby 2021 haskell 2022 typescript 2023 typescript
Français