Gonzalo Espinoza Graham 🏴☠️
@geepytee
Followers
4K
Following
39K
Media
825
Statuses
9K
on sabbatical @uwaterloo before it was cool categorized as a top 20% male by @levelsio I like to see friends win
SF
Joined March 2013
GPT-4V + TTS = AI Sports narrator 🪄⚽️. Passed every frame of a football video to gpt-4-vision-preview, and with some simple prompting asked to generate a narration. No edits, this is as it came out from the model (aka can be SO MUCH BETTER)
182
644
4K
@robkhenderson Startup idea: CVS but with memberships so stuff doesn't need to be locked up.
45
31
2K
@drawbrandondraw It's crazy that you can just 'download' cancer. Truly living in the future.
11
19
1K
@soren_iverson This is genius. And if you are awake, you get to push a button to play loud noises from the sleepers' phones!.
0
36
1K
@stargazermellu Wait is this why there's always people eating in their cars? . Never understood this part of the american experience.
5
18
804
Building hardware again, starting with a robot dog. Full 3D-printed, open-source design, with an ML controller inspired by Disney's bipedal animated robot. Documenting the entire build on stream.
not a lot of long-form hardware engineering content out there . been liking this series from @geepytee . it’s cool to see how a pro builds things, from start to finish
14
24
483
I see a lot founders trim their absurdly ambitious ideas when pitching to investors. This is a mistake. You want to be Sam Altman telling a room full of VCs with a straight face that he will take billions of their money, build AGI, and then ask it how to generate a return
31
34
408
1
0
326
@johnkrausphotos You know you're living in the future when it's news that a rocket had an accident on its 23rd landing in a row.
6
15
327
@SteveMoraco Video is 1131 frames long, but only passed every 10th frame all of them together in a dictionary (I think this only counts as 1 request but a ton of tokens?). Not cheap (~$30). Here's the full code: Based on OAI's cookbook
9
27
289
@TrungTPhan @gdb @sama The real shocker here is that after $10B from Microsoft, OpenAI still refuses to use Teams.
13
11
198
Friend in YC is building a humanoid robot. Hosting an unveil next Friday where it will take its first steps 👣. DM for details if you want to come see :)
12
3
162
Built an OpenGrip claw tonight. This will eventually turn into a helping hand in my kitchen (ultrasonic sensor to detect objects in front of it and grab them). Or maybe a full ALOHA teleop arm?? 👀 . Really cool to experience the state of open source robotics, future is bright
some friends working in robotics expressed interest in a gripper I was designing for my senior capstone project, so I've decided to make it open source! it's extremely simple and cheap, but doesn't sacrifice on performance 🦾. check it out:.
11
9
131
First day at Waterloo I joined FSAE. Nothing beats building a race car with your friends, why would you want to do anything else?. 2 months in, I realized they didn't let freshman build anything fun (my project was the headrest in the car). I went to the dean's office and pitched
Two of SpaceX's senior execs Mark Juncosa and Bill Riley are Cornell grads who credit their engineering chops to the Formula1 project team (FSAE) in college. Join clubs in college to actually building stuff together, not the "consulting club" which is just college kids chatting.
6
3
122
Got access to ChatGPT vision, it changes everything. Fed it a Safeway flyer, asked it to come up with the cheapest meal possible while following nutritional guidelines. BOOM
4
13
117
@stillgray You are wrong. I can’t speak for the rest of the world but Canada has the same, if not more strict, mask and distancing laws.
3
0
90
In 2021 you could get anything delivered by robots in Waterloo. This was a service @WesleyYue and I launched called WaterlooRobotDelivery dot com. Someone recorded our robot mid delivery and it went viral!
5
4
101
OpenAI TTS + Real Time Streaming = Audio in 0.6secs 🪄🔊. Instead of waiting for the voice MP3s to generate, I tried the new real time streaming feature. 1.5 second of waiting for first audio, down to 0.6 seconds 👀 Excuse my heavy breathing, got excited. Data in thread!
14
5
94
@soren_iverson These idea tweets started as jokes, but this one is genius and genuinely useful.
4
0
92
The first ever humanoid robots olympics went down this weekend at the @kscalelabs hackathon. We went undefeated in boxing ;)
6
7
86
Today @WesleyYue and I are launching Double 🚀, a new AI tool that cleans, enriches, and qualifies your leads!. It's available to everyone for free at Here're 4 examples to show you how Double will automate your prospecting process 🧵
11
15
70
@montana_skeptic @Tesla What's with the hate? If you are not going to help, at least shut up and move aside.
2
0
69
@jackson_llk Lo puse en español al comienzo pero tenia que estar en ingles o sino no hiba a ser viral.
4
0
65
@nikitabier Instagram will be 1/3 ads, 1/3 AI generated content, and 1/3 people trying to hook up.
2
0
59
Kept tinkering with OpenAI's TTS and ways to reduce first-audio latency:. GPT-3.5-Turbo responses streamed back. Chunk streamed responses into sentences and send to TTS. TTS streaming to play before file is fully gen.=.Consistent first audio in ~1sec regardless of length of msg
OpenAI TTS + Real Time Streaming = Audio in 0.6secs 🪄🔊. Instead of waiting for the voice MP3s to generate, I tried the new real time streaming feature. 1.5 second of waiting for first audio, down to 0.6 seconds 👀 Excuse my heavy breathing, got excited. Data in thread!
7
9
52
1
6
36
They say you should make something people want 🧡. Since launching Claude 3 Opus in we went from 100 -> 10k+ installs in 3 weeks. Now trending on the VS Marketplace and subscription revenue growing quick. 100X'd our user base for identifying a new LLM
New Claude 3 model claims to outperform GPT-4. Want to try it? Now available to chat with on free for the time being. Early tests are very promising!.
11
6
48
@ashleevance Crazy to me that the new CEO didn't ask before taking the role, this would be my 1st question.
2
0
45
@kyleturman @tiffani I think the idea is that by default everyone is contributing through taxes, and failure of the local government to provide the minimum services doesn't foster an environment for building and contribution. Individuals cannot fix governments as it stands.
2
1
45
3
0
46
Double's merch dropping soon.
“It’s actually at $3K MRR now” @geepytee . Amazing energy at the @ycombinator investor reception, feels like OG in-person demo day
4
0
42
New Claude 3 model claims to outperform GPT-4. Want to try it? Now available to chat with on free for the time being. Early tests are very promising!.
Today, we're announcing Claude 3, our next generation of AI models. The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision.
8
5
41
@TenreiroDaniel After listening to this pod I went and ran my first LLM *locally*. You can already tell they are going to try to regulate this to hell.
2
0
35
@AmandaMGoetz I think the bare minimum is:. negative AIDS test.negative COVID test.full credit report.police check.college diploma.proof of income.
4
0
33
@exec_sum For the people asking where this is from, Axios puts these out every year by @StefWKight . 2022: 2021:
1
3
33
@paulg @_realprimetime @garybasin @MsMelChen Paul Graham reminding us that even our idols are human.
0
0
33
@finbarrtimbers @ArtBasler There's an alternative world where, instead, Zuck spends billions a year on the OpenAI API.
0
1
31
I want to upload all my emails, conversations, messages, vlogs, etc to GPT-3. Most people have a voice inside their head they listen/talk to. Imagine if you could text that voice. Text yourself and get a coherent reply from yourself. Have a two-sided conversation with yourself.
2
2
25
Llama 3 is now available at you can try it for free right now 🫡
8
0
27