Henry Semakula

BSc. Information Technology from Makerere University.
MSc. Embedded and Mobile Systems from NM-AIST.
Interested in AI | Deep Learning | Machine Learning.

[Turing Post @TheTuringPost]

Must-read AI research of the week:

▪️ OpenClaw-RL
▪️ Meta-Reinforcement Learning with Self-Reflection for Agentic Search
▪️ Agentic Critical Training
▪️ Video-Based Reward Modeling for Computer-Use Agents
▪️ AutoResearch-RL
▪️ Neural Thickets
▪️ Training Language Models via Neural Cellular Automata
▪️ The Curse and Blessing of Mean Bias in FP4-Quantized LLM Training
▪️ Lost in Backpropagation: The LM Head is a Gradient Bottleneck
▪️ IndexCache
▪️ Attention Residuals
▪️ REMIX: Reinforcement Routing for Mixtures of LoRAs in LLM Finetuning
▪️ Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections
▪️ Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs
▪️ How Far Can Unsupervised RLVR Scale LLM Training?
▪️ Examining Reasoning LLMs-as-Judges in Non-Verifiable LLM Post-Training
▪️ Reading, Not Thinking: Understanding and Bridging the Modality Gap When Text Becomes Pixels in Multimodal LLMs
▪️ Scale Space Diffusion

Find the full list and the main AI news and updates from NVIDIA GTC here: https://t.co/T985DbaCvR

[Turing Post @TheTuringPost]

5 new LoRA variants that are changing Fine-Tuning

▪️ Doc-to-LoRA
▪️ Text-to-LoRA
▪️ Mixture of Adapters (MoA)
▪️ LoRA-Sqeeze
▪️ Kron-LoRA

Here’s how they work and what stack they form → https://t.co/0fjxDD2syS

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

Most health AI systems are trained on data that doesn’t reflect how people in many African contexts actually ask questions about sexual health across languages, cultures, and levels of digital access.

That gap over time creates real safety risks.

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

To address this, we have spent the last two years crowdsourcing real STI-related questions from all around Africa, capturing how people actually express concerns, uncertainty, and risk rather than how systems expect them to.

The result?

A dataset grounded in a lived context.

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

🚨ANNOUNCEMENT🚨
After 2 years of crowdsourcing real STI-related qns, we are delighted to announce that our HASH STI dataset is now publicly available for research, model evaluation, & health AI development.

🔗https://t.co/AT6lqrMAMM

But why does this dataset matter?

A thread

[Academy for Health Innovation, Uganda @TheAcademy_IDI]

Last Thursday, our medical drones team in Kalangala showcased a live drone delivery to @ProfNawangwe and the @IDIMakerere Board members.

The demonstration took place at Bufumira Island Health Centre, where drones pick up ARVs for onward delivery to the neighboring islands.

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

Through a co-creation approach, teams engage young people, health experts, and policymakers to shape an agent that is accurate, culturally relevant, multilingual, and responsive to real MSRH information needs across countries.

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

This conversational agent is being developed collaboratively with three leading institutions: University of Embu (Kenya), University of Ghana–Legon, and Addis Ababa University of Science & Technology (Ethiopia). Each brings local expertise and user insights.

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

In HASH Phase II, we are co-creating a multinational conversational agent to strengthen maternal, sexual, and reproductive health across Africa.

This next phase builds on lessons from Phase I to scale a context-aware, locally built AI tool. 🧵

[Academy for Health Innovation, Uganda @TheAcademy_IDI]

A two-day Obstetric Ultrasound AI hackathon organised by @AiHub4MSRH featured software developers & healthcare professionals from Uganda, Kenya, Nigeria, Pakistan & Nepal who worked on a challenge to predict fetal gestational age in days using the 8-blind-sweeps data set.

[Academy for Health Innovation, Uganda @TheAcademy_IDI]

From 6th to 7th November 2025, the key conversations across the two days explored AI policy and ethics, localization of AI solutions, capacity building, AI for strengthening health systems, the role of generative AI, and sustainable business models for innovation.

[Academy for Health Innovation, Uganda @TheAcademy_IDI]

The Academy cordially invites you to the public PhD Defense of our Innovate Theme Lead @AgnesBwanika

Date: 05/11/2025
Time: 09:00 AM
Venue: MakCHS, Conference Room
Join the session online: https://t.co/gpV9brhEyU

[Makerere University @Makerere]

AI in Health Africa Conference hosted by @AI_HealthLabMak
Join us for the 2nd AI in Health Africa Conference on 6th – 7th, November 2025 at Makerere University, Main Hall where experts, policymakers, researchers, and healthcare innovators will explore AI-driven solutions tailored to Africa’s unique healthcare landscape and the localization of AI in African healthcare systems.

[Academy for Health Innovation, Uganda @TheAcademy_IDI]

The Call for Speakers for the 2nd AI in Health Africa Conference is now OPEN. We invite innovators, researchers, policymakers, & health professionals to share on Scaling AI for sustainable & inclusive health systems in Africa.

Register by 30th September 🔗https://t.co/hO4qRXV0Xa

[Infectious Diseases Institute, Makerere University @IDIMakerere]

Are you passionate about improving maternal & fetal health? You've a chance to win an all-expenses-paid trip to Uganda to showcase your solution at the Africa Health AI Conference 2025.
Apply NOW: https://t.co/gKO7kxwzOD #OneIDI #AIHealth #Hackathon

[Daily Monitor @DailyMonitor]

The Personal Data Protection Office found that Google qualifies as both a data controller and collector under the Data Protection and Privacy Act; therefore, its failure to register as a data collector violated Ugandan laws
#MonitorUpdates https://t.co/7umM3pwV6j

[Academy for Health Innovation, Uganda @TheAcademy_IDI]

Our very own @MirtyneJ will deliver a keynote address at the Africa AI Health Forum in Kumasi, Ghana.

This year’s theme:
Harnessing Generative AI for Healthcare Innovations – Lessons, Challenges, and Future Directions.

To attend online, register here: https://t.co/aeXgxDsPKU

[Academy for Health Innovation, Uganda @TheAcademy_IDI]

Join our friends at @AI_HealthLabMak for an insightful conversation on aligning AI innovations with Africa's rich traditional health knowledge.

We're excited to attend & engage as we build momentum towards the upcoming AI in Health Africa Conference.
https://t.co/UU6c4s0RMl

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

Our Principal Investigator @rratanshi delivered a presentation on AI in Health equity & equality and shared the lessons learned from phase one of the HASH project at #HAICon2025 in Kathmandu, Nepal.

Follow this link to discover more about what we learned
https://t.co/eTXlcSVrcu

[Hub 4 AI in Maternal,Sexual & Reproductive Health @AiHub4MSRH]

At the launch of Phase 1, our subgrantees shared AI innovators' needs. As we close out, we're excited to reconnect & assess how well those needs were met & the challenges they still face, helping to shape the network. To get updates about the HASH network: https://t.co/hmJrXeZKoZ