🤖 International Olympiad in AI - Senegalese Delegation Selection
This article details the national selection process organized by GalsenAI to form the Senegalese delegation for the International Olympiad in Artificial Intelligence (IOAI). The process involved multiple phases across various regions, coordinated with academic inspectorates.
Key Points:
• GalsenAI organized the national selection for the IOAI.
• The selection process occurred in three phases.
• Multiple regions of Senegal participated.
• Academic inspectorates and education training facilitated the process.
🔗 Resources:
Image
Image
Image
Image
✨ AI Leadership - AV Luminary Award Winners
This article announces the recipients of the AV Luminary Award at DataHack Summit 2026. The award recognizes individuals for their contributions to AI-led transformation and enterprise-scale adoption of AI across industries.
Key Points:
• The AV Luminary Award acknowledges leaders in AI transformation.
• Winners were announced at DataHack Summit 2026.
• Abhishek Kumar and Andre Zayarni received the awards.
• The award focuses on enterprise-scale AI adoption.
🔗 Resources:
Image
🤖 LLM - Qwopus3.6-27B-Fusion Model
This article introduces Qwopus3.6-27B-Fusion, a large language model created by merging Qwen3 and Qwen3.5. It uses task vectors and DARE TIES to combine the strengths of both models for improved reasoning and code generation.
Key Points:
• Qwopus3.6-27B-Fusion merges Qwen3 and Qwen3.5 architectures.
• It uses task vectors and DARE TIES for model weight fusion.
• The model has 27B parameters and is GGUF quantized.
• It is designed for both reasoning and coding tasks.
• The fusion approach combines model weights using layer-weighted methods.
🔗 Resources:
• Model Details ↗ - Qwopus3.6-27B-Fusion model details
Image
🤖 NLP - LFM2.5-Encoder-350M
This article describes LFM2.5-Encoder-350M, a bidirectional masked language model designed for NLP tasks. The model fills in missing words by understanding both left and right context, offering a balance between performance and efficiency.
Key Points:
• LFM2.5-Encoder-350M is a bidirectional masked language model.
• It processes both left and right context to predict hidden words.
• The model has 350M parameters, balancing performance and efficiency.
• It is part of the LFM2.5 family for language understanding.
• Custom code and safetensors format simplify model integration.
🔗 Resources:
• Model Details ↗ - LFM2.5-Encoder-350M model details
Image
⭐️ Support
If you liked reading this report, please star ⭐️ this repository and follow me on Github ↗, 𝕏 (previously known as Twitter) ↗ to help others discover these resources and regular updates.
