Malaysia's Open-Source AI Opportunity
Malaysia can leverage open-source AI, like DeepSeek, to boost innovation, ensure data autonomy, and address cultural biases by localizing LLMs.
Tag: Fine-Tuning
The Deepseek-R1 Effect: Reasoning Model Innovation
Deepseek-R1 has catalyzed reasoning-enabled language model innovation, spurring replication and new approaches with data quality, RL, and training strategies.
Personalized AI: Fine-Tuning OpenAI's o4-mini
Explore OpenAI's RFT for o4-mini: Tailor AI to your enterprise's needs with powerful customization and control.
Unlocking Knowledge Distillation: AI Model Learning
Explore how knowledge distillation enables large AI models to transfer expertise to smaller, more efficient models, revolutionizing AI scalability and deployment in diverse applications.
Nvidia Surpasses DeepSeek: Open-Source Model Deep Dive
Nvidia's open-source Llama-Nemotron series outperforms DeepSeek-R1. Details of its 140,000 H100 training hours are disclosed, marking a major advance in accessible AI technology.
Microsoft's Tiny Model: Math 'Cheat Code' on 6K Samples
Microsoft's small Phi-4 reasoning models achieve impressive math performance with limited data. Surpassing larger models, they demonstrate the power of fine-tuning and reinforcement learning in AI.
Meta's LlamaCon: LLM Deep Dive & Future Insights
A comprehensive analysis of Meta's LlamaCon, exploring the present state and future trajectory of large language models, multimodal applications, and the open-source landscape.
DeepSeek: Democratizing AI with Lower Costs
DeepSeek's discounted foundation models tackle AI adoption costs, potentially revolutionizing enterprise use. This move empowers smaller companies and accelerates AI implementation across industries.
Microsoft's Phi-4-Reasoning-Plus: Compact AI Powerhouse
Microsoft's Phi-4-reasoning-plus is an open-weight language model designed for advanced reasoning tasks. It uses fine-tuning and reinforcement learning to excel in math, science, coding, and logic.
Customizing Amazon Nova for Enhanced Tool Utilization
Explore customizing Amazon Nova models via Amazon Bedrock for improved tool utilization, enhancing decision-making and agentic workflows.