We have taken our Artificial Intelligence (#AI) Forecasting System into operations today to run side by side with our physics-based weather forecasts. The #AIFS outperforms traditional models for many measures, at a fraction of energy use. Read more ➡️ https://t.co/18f61f4cve
🌍🌐 A new climate emulator for Destination Earth (#DestinE)! @predictia and partners will develop an #ML-based #climate emulator, reproducing some features of the Climate DT at a reduced computational cost. Learn more ➡️ https://t.co/twrbm9Ah0p
New #AIFS blog: we highlight that the first AIFS model weights are now available and explain how you can use them yourself ➡️ https://t.co/ECMn50kFSY
#MachineLearning
🚀 Interested in time series generation?⏲️Excited to share my @GoogleDeepMind Amsterdam student researcher project: Rolling Diffusion Models!
https://t.co/4UXB428ZYY (to appear at ICML 2024)
Thanks for the great collaboration @emiel_hoogeboom, @JonathanHeek, @TimSalimans! 🧵1/4
FlashAttention is widely used to accelerate Transformers, already making attention 4-8x faster, but has yet to take advantage of modern GPUs. We’re releasing FlashAttention-3: 1.5-2x faster on FP16, up to 740 TFLOPS on H100 (75% util), and FP8 gets close to 1.2 PFLOPS!
1/
DYK... we're #hiring a #DevOps engineer?
We're offering:
· Competitive salary.
· Permanent, full-time contract.
· Flexible work schedule.
· Hybrid position (our office sits in Santander, in Spain's northern coast).
Find out more and how to apply: https://t.co/7NPUzPM9DQ
¡Lo que hace el Gobierno de @Cantabriaes en https://t.co/y0BX0ekFei es ejemplar!
👍 Publican datos abiertos con licencias permisivas
👍 Permiten la reutilización con muy pocos límites
👍 Integran datos de otras instituciones (ayuntamientos)
👍 Explican todo claramente, sin abusar del lenguaje administrativo.
¡Bravo! 🤩
Excited to introduce Aurora: a foundation model of the atmosphere. In <1min, Aurora produces 5-day global air pollution predictions and 10-day high-resolution weather forecasts that outperform SOTA classical simulation tools and the best specialized deep learning models… 1/n
🌍 GenCast update! 🌍 New highlights include 0.25° resolution, predicting extreme weather, cyclones, and wind power production. Across all our evaluations, GenCast is better than the world’s top operational medium-range weather forecast
New paper: https://t.co/3Ixmhuo20I
🧵 1/8
📰📺📻En los últimos días, varios medios de #Cantabria y #LaRioja se han interesado por nuestro modelo de aumento de resolución (desarrollado con @IFCA_CSIC_UC) y por nuestra charla sobre usos de la #InteligenciaArtificial en meteorología y climatología. ¡Gracias por la difusión!
🗞️Nuestro compañero @mariosanta98, explicando los entresijos del revolucionario modelo climático ConvSwin2SR para la web de @unican:
https://t.co/KaMqRzUL6E [1/2]
@MstepMIS@IFCA_CSIC_UC@KITKarlsruhe 🌿Last but not least, we present another of our use cases, integrated plant protection, explained by one of our PSNC colleagues⬇️⬇️⬇️
Vision transformers need registers!
Or at least, it seems they 𝘸𝘢𝘯𝘵 some…
ViTs have artifacts in attention maps. It’s due to the model using these patches as “registers”.
Just add new tokens (“[reg]”):
- no artifacts
- interpretable attention maps 🦖
- improved performances!
👉Are you a #Climate Data Scientist?
👉Looking forward to taking part in a #HorizonEU project?
👉Want flexible working conditions?
If you ticked ✅ all three questions, we have a #job for you!
Get all the info here: https://t.co/o4AKKpIi4w
🔚The AI4EOSC platform users’ workshop finished yesterday in Bratislava.
🌉Our colleague Mario from @predictia farewells us from Bratislavský hrad with such an amazing views of the city
Excited to share @GoogleDeepMind’s newest AI model GraphCast: the most accurate 10-day global weather forecasting system in the world. GraphCast can also offer earlier warnings of extreme weather events, including the path of hurricanes. In @Science today https://t.co/iHhQeSH3js
how do you deal with CUDA on Ubuntu?
it's a nightmare, every now and then without me touching anything my env breaks
a couple of weeks one of my screen stopped working - went black
now my torch installation doesn't even recognize underlying GPUs
do you handle everything in conda or do you also have system level installation?
if latter how do you safely purge previous version and install a new one?
I saw some gists previously by people like @jeremyphoward but can't find those