Tag: Training
All the articles with the tag "Training".
Fastino Raises $17.5M to Train AI Models on Gaming GPUs
Published: at 04:13 AMFastino secured $17.5M in funding to develop AI training solutions utilizing affordable gaming GPUs. This aims to lower AI development costs, potentially democratizing AI by challenging expensive enterprise-grade GPU dominance.
China Accelerates GPU Development Amid Policy Push
Published: at 05:22 AMThe article outlines China's significant investment and policy backing for its domestic GPU development program. This aims to achieve technological self-reliance and challenge established players in the global GPU market.
Gemma 3's Quantization-Aware Training Promises Revolutionized GPU Efficiency
Published: at 03:11 AMGoogle's Gemma 3 leverages quantization-aware training to substantially improve GPU efficiency. This technique reduces hardware requirements and makes Gemma models more accessible, unlocking new deployment possibilities and increasing competitiveness.
CoreWeave Democratizes AI Training with Cloud-Based NVIDIA Grace Blackwell GPUs
Published: at 01:57 AMCoreWeave is providing cloud access to NVIDIA's Grace Blackwell GPUs, lowering the barrier for AI training and positioning itself as a vital player in AI infrastructure by offering scalable and cost-effective solutions.
System Thinking for Scaling AI Networks to One Million GPUs
Published: at 12:05 AMThe article highlights the need for system-level thinking when designing networks for million-GPU AI training, urging co-design and application-awareness to overcome limitations of traditional architectures.
Ant Group Reduces AI Training Costs by Using Domestic GPUs, Moving Away from Nvidia
Published: at 11:40 AMAnt Group has achieved a 20% reduction in AI training costs by using Chinese-made GPUs, signaling a move towards technological independence and potentially reshaping the GPU market.