Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

HOT TAKES: GPT-6 Astra killed task-level annotation. I created a highly accurate dense task annotation system from scratch, using Astra. BEST PART IS: I packaged it as a reusable LLM skill ! Link further down

Léo

2,245 subscribers

21,662 Aufrufe • vor 16 Tagen •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

23 Kommentare

Profilbild von Stocko 🦾

Stocko 🦾vor 15 Tagen

i don't like: "tend the setting omelet" "work the omelet in the pan" wtf do those even mean it also missed the seasoning step

Profilbild von paras

parasvor 16 Tagen

🤦‍♂️🤦‍♂️

Profilbild von Léo

Léovor 15 Tagen

🦾

Profilbild von Léo

Léovor 16 Tagen

Backstory: - Robot data annotation requires, beyond hand poses and so on, tasks, subtasks, and action details. - This type of annotation requires much more human hours than basic segmentation tasks, and is also more subjective. - I noticed Astra was pretty good at video understanding, so I figured there might be something worth building.

Profilbild von Léo

Léovor 16 Tagen

Long story short, the system understands: - Tasks - Subtasks - Actions - Overlap - Hand decomposition - Observed outcome and limits - Visible evidence

Profilbild von Léo

Léovor 16 Tagen

Check my live demo on 6 samples from open source datasets: It also comes with a detailed action ledger.

Profilbild von Léo

Léovor 16 Tagen

Now the main question is: now that egocentric data generation seems to get closer to full automation, is data annotation solved as well?

Profilbild von Léo

Léovor 16 Tagen

Do you want the packaged skill ? 📟 DM me, I would love to see what you build with it !

Profilbild von Léo

Léovor 16 Tagen

A pretty insightful source by @_varunnair:

Profilbild von sankalp nagaonkar

sankalp nagaonkarvor 15 Tagen

what's the $$ per video hour my guy

Profilbild von Léo

Léovor 15 Tagen

Good question. I used my OpenAI subscription, but at current API rate, it would amount to about $0.22 per video minute.

Profilbild von Senthilnathan K

Senthilnathan Kvor 16 Tagen

This isn't new; any standard off-the-shelf VLM can ace this. Astra, in fact, is pretty heavy-duty for this specific task; the hot take is that it's killing the VLAs/WAMs themselves that these annotations are being used to train

Profilbild von Léo

Léovor 15 Tagen

please show me your demo then, I would love to see

Profilbild von Nischay Joshi

Nischay Joshivor 15 Tagen

Seems pretty interesting!

Profilbild von Léo

Léovor 15 Tagen

It sure is!

Profilbild von Niko Builds Agents

Niko Builds Agentsvor 15 Tagen

packaging it as a reusable skill is the part I care about. everyone shows the one-off demo, nobody shows the thing you can rerun tomorrow. what's the skill actually wrapping, a prompt chain or a callable tool?

Profilbild von Léo

Léovor 15 Tagen

a skill is a md file with instructions

Profilbild von yongsong yang

yongsong yangvor 15 Tagen

Amazing work, Is the skill released?

Profilbild von Léo

Léovor 15 Tagen

Yes it is, DM me and I will send oyu the link

Profilbild von Andrew Lee

Andrew Leevor 16 Tagen

Is there any benchmark for the accuracy of the annotations Astra creates?

Profilbild von Léo

Léovor 15 Tagen

unfortunately no; but if can get my hands on a human annotated dataset, I could test and benchmark Astra over it

Profilbild von Prajwal Gatti

Prajwal Gattivor 15 Tagen

@andrewheejay You can try HD-EPIC? we have dense human annotations at multiple levels of activity: high level cooking recipe -> prep/step -> detailed running action annotations

Profilbild von Léo

Léovor 15 Tagen

@andrewheejay Feel free to scrape my webpage and compare to your results. Curious to see how it goes ;)

Ähnliche Videos

I recreated Townscaper with GPT-6 Astra using Three.js on Happycapy The result is surprisingly close to the original, and it took way less time than I expected 🤯 Astra is seriously good at this kind of 3D coding task.

I recreated Townscaper with GPT-6 Astra using Three.js on Happycapy The result is surprisingly close to the original, and it took way less time than I expected 🤯 Astra is seriously good at this kind of 3D coding task.

Min Zhou

10,726 Aufrufe • vor 1 Monat

I built an ancient Indian kingdom in Three.js using Gpt 6 Astra then handed the footage to Seedance 2.5 The top is the world created by Gpt 6 Astra The bottom is how AI reimagined it

I built an ancient Indian kingdom in Three.js using Gpt 6 Astra then handed the footage to Seedance 2.5 The top is the world created by Gpt 6 Astra The bottom is how AI reimagined it

Karan

18,333 Aufrufe • vor 1 Monat

GPT 6 Astra created this 3d model of my office using blender. It is basically a perfect replica. This is crazy!!

GPT 6 Astra created this 3d model of my office using blender. It is basically a perfect replica. This is crazy!!

BridgeMind

83,142 Aufrufe • vor 1 Monat

GPT Astra vs GPT-6.1 Sol vs Sonnet 5.5 vs Opus 5.5. Which AI understands 3D best? I gave all four the exact same task: build an AK-47 from scratch. Same prompt. Same environment. Which one wins?

GPT Astra vs GPT-6.1 Sol vs Sonnet 5.5 vs Opus 5.5. Which AI understands 3D best? I gave all four the exact same task: build an AK-47 from scratch. Same prompt. Same environment. Which one wins?

Luckey Faraday

64,352 Aufrufe • vor 4 Tagen

I asked GPT-6 Astra to recreate famous scenes from Zelda OoT. It did it, using the same graphics as original game.

I asked GPT-6 Astra to recreate famous scenes from Zelda OoT. It did it, using the same graphics as original game.

Kevin Ngo

14,459 Aufrufe • vor 1 Monat

I bought ChatGPT plus ($20) I used GPT-6 Astra on "MAX" I gave it the task: "Integrate Supabase and Improve App UI" It worked for 40+ minutes! YES! The task wasn't done... These are the usage stats: - 100% 5 hr limit consumed - 16% weekly limit consumed - 5 Million tokens used - 50k Output Tokens For $20, where I can barely use Opus 5 for 20 minutes... I am getting GPT-6 Astra for 40 minutes is MORE than enough. BEST value you can get! Absolutly AMAZING!

I bought ChatGPT plus ($20) I used GPT-6 Astra on "MAX" I gave it the task: "Integrate Supabase and Improve App UI" It worked for 40+ minutes! YES! The task wasn't done... These are the usage stats: - 100% 5 hr limit consumed - 16% weekly limit consumed - 5 Million tokens used - 50k Output Tokens For $20, where I can barely use Opus 5 for 20 minutes... I am getting GPT-6 Astra for 40 minutes is MORE than enough. BEST value you can get! Absolutly AMAZING!

shownotover

198,946 Aufrufe • vor 1 Monat

People watching GPT-6 Astra demos on X may not realise how slow it actually is in real time. Here’s a rollout of GPT-6 Astra controlling the MolmoAct2 YAMs to complete this task in real-time. There is potential, but i don’t think it’s practical for real-world robotics deployment yet. I applaud your patience if you make it through the whole video.

People watching GPT-6 Astra demos on X may not realise how slow it actually is in real time. Here’s a rollout of GPT-6 Astra controlling the MolmoAct2 YAMs to complete this task in real-time. There is potential, but i don’t think it’s practical for real-world robotics deployment yet. I applaud your patience if you make it through the whole video.

Jiafei Duan

61,804 Aufrufe • vor 13 Tagen

I asked Opus 5.5, GPT-6 Astra, GPT-6 Sol, Kimi K3 to build the Eiffel Tower in Three.js. Opus 5.5: $8.95 | 10 mins GPT-6 Astra: $7.45 | 8 mins GPT-6 Sol: $3.90 | 5 mins Kimi K3: $4.04 | 6 mins Opus 5.5 really impressed me. The frontend and 3D detailing are seriously good, and I preferred it over Astra for this task. GPT-6 Sol was a bit disappointing, while Kimi K3 looks really good for the price. It delivered a strong result at almost half the cost. Kimi K3.1 is going to be a monster for sure. Overall, Anthropic made a great comeback with Opus 5.5.

I asked Opus 5.5, GPT-6 Astra, GPT-6 Sol, Kimi K3 to build the Eiffel Tower in Three.js. Opus 5.5: $8.95 | 10 mins GPT-6 Astra: $7.45 | 8 mins GPT-6 Sol: $3.90 | 5 mins Kimi K3: $4.04 | 6 mins Opus 5.5 really impressed me. The frontend and 3D detailing are seriously good, and I preferred it over Astra for this task. GPT-6 Sol was a bit disappointing, while Kimi K3 looks really good for the price. It delivered a strong result at almost half the cost. Kimi K3.1 is going to be a monster for sure. Overall, Anthropic made a great comeback with Opus 5.5.

Bhavy☄️

167,004 Aufrufe • vor 16 Tagen

GPT-6 Astra is absolutely INSANE. It created the old CONTRA in zero-shot in 30 mins. Just like I remember from childhood. So COOL.

GPT-6 Astra is absolutely INSANE. It created the old CONTRA in zero-shot in 30 mins. Just like I remember from childhood. So COOL.

Shubham Saboo

32,568 Aufrufe • vor 1 Monat

People use Jev to pick a model before a task. I made it change GPT-6's reasoning effort inside Codex DURING the task. More thinking when stuck. Less for routine steps. 50% lower Astra costs in my tests. Faster runs, without breaking prompt caching.

People use Jev to pick a model before a task. I made it change GPT-6's reasoning effort inside Codex DURING the task. More thinking when stuck. Less for routine steps. 50% lower Astra costs in my tests. Faster runs, without breaking prompt caching.

vechen

636,140 Aufrufe • vor 19 Tagen

GPT-6 Astra just surpassed CapCut. It's world class at detecting 100% of my bad takes and analyzing the best areas to put text so it doesn't cover my face. i can also watch it do the entire edit here. This was a 1-shot pass it created, all on my phone during a plane take-off. One shotted by GPT-6 Astra and Ody.

GPT-6 Astra just surpassed CapCut. It's world class at detecting 100% of my bad takes and analyzing the best areas to put text so it doesn't cover my face. i can also watch it do the entire edit here. This was a 1-shot pass it created, all on my phone during a plane take-off. One shotted by GPT-6 Astra and Ody.

Kaito | 海斗

27,158 Aufrufe • vor 1 Monat

My balance went down as I was recording a YouTube video (at 0:01), I have no idea what's going on anymore with the budget limits. Using GPT-6 Astra in Codex.

My balance went down as I was recording a YouTube video (at 0:01), I have no idea what's going on anymore with the budget limits. Using GPT-6 Astra in Codex.

Wesley

19,314 Aufrufe • vor 1 Monat

I compared GPT-6.1 Sol and GPT-6 Sol on a design task of making a clean looking landing page. 6.1 Sol looks a lot cleaner here, and its closer to Astra than 6 Sol, its a step up in a good direction! It was also a lot cheaper than 6 Sol, which is great

I compared GPT-6.1 Sol and GPT-6 Sol on a design task of making a clean looking landing page. 6.1 Sol looks a lot cleaner here, and its closer to Astra than 6 Sol, its a step up in a good direction! It was also a lot cheaper than 6 Sol, which is great

bluedev

14,089 Aufrufe • vor 10 Tagen

This is sick! I built the entire city of Gurgaon in 3D using GPT-6 Astra I saw everyone building all sorts of cool demos with Astra, so I decided to give it a shot and needless to say, I was not dissapointed

This is sick! I built the entire city of Gurgaon in 3D using GPT-6 Astra I saw everyone building all sorts of cool demos with Astra, so I decided to give it a shot and needless to say, I was not dissapointed

Yajat Gulati

50,223 Aufrufe • vor 1 Monat

GPT-6 Astra built this iOS app in 13m 3s using my AppCreator skill. I want to see my weekly mileage, not averages. Apple Health app doesn't make it easy. So I gave this prompt to Astra, and this is the result. No other model has made it work like this from one prompt. OpenAI OpenAIDevs

GPT-6 Astra built this iOS app in 13m 3s using my AppCreator skill. I want to see my weekly mileage, not averages. Apple Health app doesn't make it easy. So I gave this prompt to Astra, and this is the result. No other model has made it work like this from one prompt. OpenAI OpenAIDevs

Paul Solt

210,588 Aufrufe • vor 1 Monat

Jev x GPT-6 Astra x H3 Max I created a complex decision making game with Astra. It is about keeping the city on a whale alive with decisions. - GPT 6 Astra designed the world - Jev chose the actions - H3 Max Turbo on fal turned one decision from each round into video All of the Jev decision making and video generations took only 5 minutes for 264 clips

Jev x GPT-6 Astra x H3 Max I created a complex decision making game with Astra. It is about keeping the city on a whale alive with decisions. - GPT 6 Astra designed the world - Jev chose the actions - H3 Max Turbo on fal turned one decision from each round into video All of the Jev decision making and video generations took only 5 minutes for 264 clips

gokaygokay

21,821 Aufrufe • vor 21 Tagen

I think Astra is changing editing in a pretty subtle way. Instead of giving Astra one huge instruction — “make my video” — I can use GPT-6 Astra to create specific, highly customized pieces of the edit, then let CapCut PC bring everything together. That sounds simple, but it changes the whole workflow. #GPT6Astra #CapCutPC

I think Astra is changing editing in a pretty subtle way. Instead of giving Astra one huge instruction — “make my video” — I can use GPT-6 Astra to create specific, highly customized pieces of the edit, then let CapCut PC bring everything together. That sounds simple, but it changes the whole workflow. #GPT6Astra #CapCutPC

SARAH

121,826 Aufrufe • vor 28 Tagen

Gave GPT-6 Astra the task to plug in an ethernet cable. It did it after 2 hours and 4 human prompt interventions. Being able to complete a task with tight tolerance and deformable objects is impressive. But I think speed is still a big problem with agentic robotic systems, since deep reasoning via 3D programs is an integral part of this pipeline, and it’s not obvious to me how that can be distilled/accelerated to a level comparable to VLAs.

Gave GPT-6 Astra the task to plug in an ethernet cable. It did it after 2 hours and 4 human prompt interventions. Being able to complete a task with tight tolerance and deformable objects is impressive. But I think speed is still a big problem with agentic robotic systems, since deep reasoning via 3D programs is an integral part of this pipeline, and it’s not obvious to me how that can be distilled/accelerated to a level comparable to VLAs.

Kaifeng Zhang

17,994 Aufrufe • vor 25 Tagen

This time I had GPT 6 Astra try to break the world record using a 5x5 Rubik's Cube... The (human) world record for a 5x5 is 29.49s which Astra just barely fell short of while using computer use. I still don't think people realize how insane that is.

This time I had GPT 6 Astra try to break the world record using a 5x5 Rubik's Cube... The (human) world record for a 5x5 is 29.49s which Astra just barely fell short of while using computer use. I still don't think people realize how insane that is.

Layton Gott

64,204 Aufrufe • vor 1 Monat

JELLY PHYSICS + FUR is an absolute cheat code 😭 I built this tiny furry monster jumping game with GPT-6 Astra on Happycapy , and the interaction feels ridiculously good. GPT-6 Astra is getting scary good at this kind of interactive 3D physics. I dropped the game link in the comments. Give it a try 👇

JELLY PHYSICS + FUR is an absolute cheat code 😭 I built this tiny furry monster jumping game with GPT-6 Astra on Happycapy , and the interaction feels ridiculously good. GPT-6 Astra is getting scary good at this kind of interactive 3D physics. I dropped the game link in the comments. Give it a try 👇

Min Zhou

11,855 Aufrufe • vor 28 Tagen