AI Educator | Tech Simplified | Breaking down tools & real use cases | Making AI simple & practical | DM for Collaboration | lunatech098@gmail.com
Argentina
Joined October 2024
- Tweets2.6K
- Following338
- Followers26.4K
- Likes2.3K
Luna Tech retweeted
Open book exams test whether you can find the answer. Closed book exams test whether you actually learned it.
Most AI agents today are taking the open book version. They lean on their notes and their context, then forget everything the moment the window clears.
Continual ARC flips that. It closes the book and asks the real question: can the agent handle something new, on its own, without the notes?
That is the test that actually matters. Worth a read.
🤝 Paid partnership
OpenClaw, Hermes Agent and Prime Agent don't meaningfully learn from experience.
On Continual-ARC, our new benchmark at @letsunifyai, none of them were able to memorize anything of value and none were able to generalize to new tasks. Training a small 9B model's weights did both.
Tencent released HunyuanVideo 1.5: T2V and image-to-video weights on Hugging Face, 720p clips at 5s or 10s per the model card, plus inference code. For short video, iterate locally instead of paying per render to a closed API. Check the license. huggingface.co/tencent/Hunyu…
1
1
25
At the ElevenLabs Summit, @mati previewed Interaction Engine, a new system that composes and orchestrates the models behind conversational agents.
It uses less expensive models for simple requests and frontier models for complex ones like refunds, so you get the right outcome without overspending on tokens.
OpenClaw, Hermes Agent and Prime Agent don't meaningfully learn from experience.
On Continual-ARC, our new benchmark at @letsunifyai, none of them were able to memorize anything of value and none were able to generalize to new tasks. Training a small 9B model's weights did both.
Luna Tech retweeted
The real test of a learning agent is not what it remembers. It is what it can do with a task it has never seen before.
That is the bar. Any agent that claims it learned something should be able to prove it on a problem it was not trained on.
This is exactly what Continual ARC goes after, and the results are worth your attention.
🤝 Paid partnership
OpenClaw, Hermes Agent and Prime Agent don't meaningfully learn from experience.
On Continual-ARC, our new benchmark at @letsunifyai, none of them were able to memorize anything of value and none were able to generalize to new tasks. Training a small 9B model's weights did both.
The real test of a learning agent is not what it remembers. It is what it can do with a task it has never seen before.
That is the bar. Any agent that claims it learned something should be able to prove it on a problem it was not trained on.
This is exactly what Continual ARC goes after, and the results are worth your attention.
🤝 Paid partnership
OpenClaw, Hermes Agent and Prime Agent don't meaningfully learn from experience.
On Continual-ARC, our new benchmark at @letsunifyai, none of them were able to memorize anything of value and none were able to generalize to new tasks. Training a small 9B model's weights did both.
Luna Tech retweeted
OpenClaw, Hermes Agent and Prime Agent don't meaningfully learn from experience.
On Continual-ARC, our new benchmark at @letsunifyai, none of them were able to memorize anything of value and none were able to generalize to new tasks. Training a small 9B model's weights did both.
🚨💥 $1,000 BTC GIVEAWAY 💥🚨
🏆 5 WINNERS × $200 in Bitcoin each
⏳ 5 days. That’s it.
To enter (all 3 required):
1️⃣ Follow @ElizabethAI_1
2️⃣ Retweet this post 🔁
3️⃣ Tag 2 friends 👥
Winners picked randomly and announced live. 🎲
Don’t scroll past this one. 👀🔥
#Bitcoin #BTC #CryptoGiveaway #Giveaway
Luna Tech retweeted
🚨💥 $1,000 BTC GIVEAWAY 💥🚨
🏆 5 WINNERS × $200 in Bitcoin each
⏳ 5 days. That’s it.
To enter (all 3 required):
1️⃣ Follow @ElizabethAI_1
2️⃣ Retweet this post 🔁
3️⃣ Tag 2 friends 👥
Winners picked randomly and announced live. 🎲
Don’t scroll past this one. 👀🔥
#Bitcoin #BTC #CryptoGiveaway #Giveaway
Kimi K2 Thinking is on Hugging Face with open weights: Moonshot's reasoning-tuned variant for multi-step agentic work and tool calls. Takeaway: you can self-host an open reasoning model tuned for tool use, no per-task API bill. huggingface.co/moonshotai/Ki…
1
2
16
One stage, one room, 26 camera angles. Our AI band's video is built like a live set: the only thing that changes is the camera. The shot plan covers 277.974s against a 277.974s master, to the millisecond. (6/8)
1
1
12
Two masters vanished. Our scene docs said "duration unknown, do not invent a runtime." Then they came back: 164.88s and 175.15s, checksummed. An AI label that keeps its own failures on the record. (7/8)
1
1
22
Your LiveAvatar just leveled up.
Any frame: close-up, half-body, full-body
Every feeling: 6 emotions
Now in 1080p
All from a single image.
Available now 👇
Luna Tech retweeted
He chose the wrong girl to mess with, and now he’ll never forget what happened.. 🎬
Everyone is racing to build an AI tutor.
Paradigm built something stranger. An AI that watches you learn and jumps in before you even ask.
Meet Clover the first ambient learning agent.
Most AI sits there waiting for the perfect prompt. Clover watches your canvas. You stall on a slide, cross out a line, freeze mid-problem, and the hint is already there.
I dropped in a lecture slide I was stuck on. It zoomed into the exact step that was breaking, traced it one move at a time, and circled where it fell apart. No prompt. It just noticed.
This is the part nobody is ready for.
If a school like this handed you a certificate one day, would you recognize it? What would you need to verify before you did?
That question is coming faster than people think.
paradigm.study
@ParadigmStudy
🤝 Paid partnership
