Neo-Arts | Meta-Explorer I defend my idea-l-s, if they offend you, it means my defenses are working #ClimateActionNow
eKapa, South Africa
Joined July 2021
- Tweets103K
- Following2.9K
- Followers1.3K
- Likes283K
Ncontsi retweeted
Swedish activist Greta Thunberg arrived at COP31, her first time attending the climate summit since 2021. She said the impact of the talks could have 'life or death consequences on 'countless people' reut.rs/4BS6Eym
Ncontsi retweeted
If you're building LLM apps, they'll need more than basic monitoring.
In this tutorial, Jessica teaches you how to build end-to-end LLM observability in FastAPI with OpenTelemetry.
You'll trace retrieval, prompt building, model calls, token usage, cost, and evaluation signals.
freecodecamp.org/news/build-…
Ncontsi retweeted
Five human demos. Under five minutes. A robot learns a brand-new task, from recording to deployment.
That's Steertuning from NYU: 71% average success across 12 new tasks, including kitchens in new homes.
Ncontsi retweeted
They say China's the biggest threat in AI, the reality is China's cash may be holding up the entire market.
Really interesting. Have folks come across anecdotal instances in which a lab's own agent/model becomes "frustrated" with its own harness? All 3 are from the same lab, but I took notes on its reasoning at random points during turns, and it was genuinely "frustrated" a few times.
If your agent has been working on a long horizon task unsupervised for ~3 days, how do you get around reviewing its extensive reasoning traces to ensure you understand how & why it made key decisions along the way? It's a lot of reading to get through, or do folks just skip that?
5
Ncontsi retweeted
In 2015, scientists expected the world would pass the 1.5C threshold in 2042–27 years down the road. Now they expect the threshold will be passed in 2029. In eleven years we’ve accelerated warming so much that we’ve shaved thirteen years off the timeline.
In 2015, the projected date for reaching 1.5°C above pre-industrial levels was 2042, based on ERA5 data. Today, our C3S global temperature-trend-monitor anticipates the threshold breach within this decade.
Read the article: climate.copernicus.eu/how-fa…
#CopernicusClimate
Ncontsi retweeted
A 27-year-old Djavan made a brief uncredited appearance as a supporting actor in the original 1976 TV Globo telenovela Anjo Mau,
Ncontsi retweeted
Science gets interesting when the answer refuses to settle down.
Entanglements: The Undark Podcast follows the questions that remain open even after the research has piled up.
Catch up on the latest conversations from Undark!
🔊 Available on Apple Podcasts, Spotify, Pandora, and SoundCloud.
🔗: undark.org/podcast/
Ncontsi retweeted
I agree with Queen B. Gayton McKenzie needs to resign. I refuse to accept that South Africa can spend R31 million on a World Cup delegation while our own swimmers must find R55,000 to represent the country. If this is his idea of prioritising sport, we've got a serious problem.
Ncontsi retweeted
A year and a half of searching, waiting, and holding on to a thread of hope ended with finding the remains of the child Ibrahim beneath the rubble.
Ncontsi retweeted
From being a Columbia University encampment organizer to becoming the next Congresswoman for Columbia University's district. 💗
Ncontsi retweeted
The hardest document in life insurance isn’t the application.
It’s the medical records packet that comes with it. Hundreds of pages of faxed APSs, lab reports and clinic notes from different providers, in no particular order.
Underwriters need a handful of facts from it. If you miss one, the policy gets priced on an incomplete picture.
On Wednesday, Nov 18 at 9 AM PT, we’ll build the whole workflow live with LlamaParse and show you how to set up your own. It’s free, and you’ll leave with something you can run on your next packet:
1/ Splitting the packet into individual documents by provider and date
2/ Parsing scanned and faxed pages, from handwritten notes to lab tables
3/ Extracting diagnoses, medications and lab values with page citations
4/ Indexing the full record so you can ask follow-up questions
Instead of digging through hundreds of pages or rereading the entire file, you only check the cited page and move on.
RSVP here: watch.getcontrast.io/registe…
Ncontsi retweeted
Great practical paper on optimizing harness costs.
For lower agent costs, you want to change the harness and keep the model.
UC Berkeley researchers found that harness changes such as prompts and tools barely move accuracy, but they change cost by up to 2.6x on the same model and task.
They analyzed agent traces and found five ways to cut cost. Each one reduces how much context an LLM call reads or how many calls a task needs.
Then they built Agency, a layer that sits between existing harnesses like Claude Code and Codex and the model. An optimization written once works across harnesses.
Five proof-of-concept techniques cut cost by up to 35% with marginal accuracy loss.
If your agent bill keeps growing, look at repeated context reads and extra tool calls before switching models.
arxiv.org/abs/2611.12122
Chat with Paper: academy.dair.ai/papers/syste…
Ncontsi retweeted
Unemployment Rate in South Africa | Q3 2026 (Stats SA)
National: 33.7% (+0,1 pp)
🟨 Johannesburg: 34.88% (−0.99 pp)
🟦 Cape Town: 22.21% (+0.33 pp)
🟩 Tshwane: 34.47% (+0.22 pp)
🟨 Ekurhuleni: 37.76% (+3.15 pp)
🟨 eThekwini: 22.20% (+0.99 pp)
Ncontsi retweeted
THE FIFTH DIMENSION — EPISODE 2 🎙️
Our first special guest: Ben Shapiro.
Three microphones. Two brain cells. Absolutely no follow-up questions.
Ncontsi retweeted
NEW Trump was happy to support the duty-free trade deal renewal (AGOA) for Africa --- as long as South Africa and Rwanda were cut out of it. @dan_m_ford
responsiblestatecraft.org/tr…
