PhD Researcher in Biostatistics-Machine Learning @Ucalgary || Graduate Teaching Assistant @Faculty of medicine and Dept of Mathematics and Statistics @Ucalgary
Call for Tutorials & Skill Sessions: Deep Learning Indaba 2026 ✨
Under this year’s theme, "Sovereign Intelligence: Africa’s Path in a Frontier AI World," we are officially opening our first-ever open call for Tutorials and Skill Sessions. We are looking for experts and educators to empower the African community with the tools to build, control, and deploy AI.
What are we looking for?
🔹Tutorials (2hrs): Foundational, lecture-style sessions (e.g., GNNs, Climate Modelling, Agentic AI).
🔹Skill Sessions (1hr): Hands-on "how-to" demos (e.g., shipping production code, research writing).
Whether you are teaching core fundamentals or exploring technological sovereignty through open-source stacks, we want to hear from you.
🗓️ Deadline: 8th March 2026 (AOE)
📍 Where: In-person in Nigeria (with post-Indaba online opportunities!)
Apply here: https://t.co/Vub672VO4q
Let’s build a foundation for African AI agency together!
#DLI2026 #Indaba2026
Meet the Captains steering the ship to Nigeria! 🇳🇬
The Deep Learning Indaba is community-led, and every year, a dedicated team steps up to make the magic happen. We are thrilled to introduce the General Chairs for #Indaba2026!
Bringing together a powerhouse of experience across research, industry, and the local ecosystem:
- @AishaAlaagib – Applied Researcher at Thakaa
- @ankahira – AI Engineer at AstraZeneca
- Chinazo Anebelundu – Director, AI Solutions Delivery at @dsn_ai_network
They are already hard at work curating a world-class programme and ensuring that our time in Nigeria is unforgettable.
Let’s give them a warm Indaba welcome!
#DLI2026 #Indaba2026
Meet the Captains steering the ship to Nigeria! 🇳🇬
The Deep Learning Indaba is community-led, and every year, a dedicated team steps up to make the magic happen. We are thrilled to introduce the General Chairs for #Indaba2026!
Bringing together a powerhouse of experience across research, industry, and the local ecosystem:
- @AishaAlaagib – Applied Researcher at Thakaa
- @ankahira – AI Engineer at AstraZeneca
- Chinazo Anebelundu – Director, AI Solutions Delivery at @dsn_ai_network
They are already hard at work curating a world-class programme and ensuring that our time in Nigeria is unforgettable.
Let’s give them a warm Indaba welcome!
#DLI2026 #Indaba2026
Last week, Humanity’s Last Exam was published in @Nature.
In just over a year, model scores on HLE have risen from under 5% to nearly 40%.
Thank you to @scale_AI and the 1000+ HLE co-authors for helping policymakers and the public track these rapid advances in AI capabilities.
Stanford just dropped their full LLM course on YouTube.
9 lectures.
Completely Free.
Real curriculum-level depth.
CME 295: Transformers & Large Language Models
This isn’t:
• a hype tutorial
• a prompt-engineering hack
• a tech influencer hot take
It’s Stanford’s Autumn 2025 course.
They cover: Transformers from first principles
Tokenization, attention, positional embeddings
Decoding, MoE, scaling laws
LoRA, RLHF, fine-tuning
RAG, tool calling, evaluation
RoPE, quantization, optimization tricks
This is foundation-level AI knowledge.
The kind that actually gets you ahead.
If you’re serious about learning AI:
👉 bookmark this
👉 repost for later
👉 stop doomscrolling and build
Playlist link: https://t.co/6rEBdY9tIU
VL-JEPA is out! A non-generative vision-language model, based on JEPA. Different from typical data-space autoregressive VLMs, VL-JEPA is trained to predict continuous embeddings in the latent space. (1/N)
Announcing the ICML 2026 policy for LLMs in reviewing! Reviewers and authors both pick either conservative or permissive LLM use, and will be matched accordingly. Importantly: authors on papers who choose conservative must obey the conservative policy as reviewers.
Multi-Agent Evolve is now fully open-source 🚀
With our codebase, you can pick your favorite LLM checkpoint and let it self-evolve, WITHOUT external supervision
💻Code:
https://t.co/GSetfobMwE
🤗Model Checkpoints:
https://t.co/Bz583Rg1s0
Feedback and contributions are welcome!
A NEW PAPER FROM YANN LECUN:
LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics
This could be one of LeCun's last papers at Meta (lol), but it's a really interesting one I think. Quick summary:
Yann LeCun's big idea is JEPA, a self-supervised learning method. However, there are various failure modes of this approach, so training strong JEPA models is very brittle, unstable, and quite difficult. So overall JEPA has seen little adoption in practice.
This paper tries to directly address this, making specific design decisions that improve training stability.
The authors identify the isotropic Gaussian as the optimal distribution that JEPA models’ embeddings should follow and design the Sketched Isotropic Gaussian Regularization (SICReg) to constrain embeddings to reach that ideal distribution. This forms the LeJEPA framework, which can be implemented in ~50 lines of code.
On empirical tests, the authors demonstrate stability of training across hyperparameters, architectures, and datasets.
A result particularly interesting to me however is that training a LeJEPA model from scratch directly on the downstream dataset outperforms finetuning a DINOv2/v3 model on the dataset!
LeJEPA: a novel pretraining paradigm free of the (many) heuristics we relied on (stop-grad, teacher, ...)
- 60+ arch., up to 2B params
- 10+ datasets
- in-domain training (>DINOv3)
- corr(train loss, test perf)=95%
Paper: https://t.co/NpfB9G1pOP
Code: https://t.co/BsK5wmNEHc
🔥Excited to share our latest work on soft-thinking LLM reasoning, SofT-GRPO.
💊 The soft-thinking paradigm for LLM reasoning can outperform the conventional discrete-token CoT reasoning in some scenarios. However, while the discrete-token CoT reasoning pattern can be reinforced through policy optimization algorithms such as GRPO, extending the soft-thinking pattern with RL remains challenging.
💡We present The First Powerful RLVR Algorithm for Soft-Thinking Reasoning: SofT-GRPO, which integrates the Gumbel-Softmax technique into the group rollout process and uses the Gumbel reparametrization trick for gradient estimation.
📝 SofT-GRPO can outperform discrete-token GRPO on Pass@16 (+1.80%), Pass@32 (+2.19%), and demonstrate a stable lead on Pass@1 (+0.13%).
📝 SofT-GRPO can also demonstrate better out-of-domain generalization ability to coding and scientific reasoning.
📖Paper at https://t.co/z3b0cMWBAo
⭐Code at https://t.co/iEBJH1gqsP
🤗Huggingface for weights https://t.co/GB2rsFTyO1
Please help spread the word — share this link with your network, communities, and social groups so more people can experience accessible, intelligent healthcare. Let’s shape the future of digital health together.
#DigitalHealth#HealthcareInnovation#MentalHealth#healthTech
Amazing innovation by VirtuClinic!
Their Digital Health Advisors (DHAs) are friendly AI assistants that instantly connect people to trusted health and mental-health support — in seconds.
You can try one here 👉 https://t.co/EIQ1xyCqQi
. No more long waits. No more multiple providers.
. Check them out: https://t.co/tP2TzSKG1e
Their new Digital Health Advisors are next-level — try them out and complete the 1-minute feedback survey to help make them even better. https://t.co/xlBsaNCY9g
They offer same-day online medical and mental-health support from trusted clinicians, without the hassle of repeating your medical history every time. Their centralized records ensure continuity of care, privacy, and respect for your time.
About VirtuClinic:
VirtuClinic is a full-service medical clinic providing a complete spectrum of physical and mental healthcare for individuals, families, and businesses — available virtually across Alberta.