AI Briefing#scaling
The latest AI and developer news about #scaling, with the original source and a short summary.
The latest AI and developer news about #scaling, with the original source and a short summary.
We confirmed that the optimal number of domain data repetitions increases with model size during LLM scaling.
Xiaomi validated VLA scaling using 100,000 hours of manipulation data collected without robots.
Explorative Modeling adds an exploration axis to generative models to boost efficiency.
DeepSeek founder Liang Wenfeng told investors in a meeting that achieving AGI, not monetization, is the company's top priority.
Speechify adopted Vercel and Next.js, cutting infrastructure costs by 50% and achieving 99.99% uptime.
New research shows that with sufficient compute power, training large models on low-quality data without filtering is possible.
Growing a 32-expert MoE to 64 experts mid-training cut GPU hours by about 32%.
By expanding the PVC first and then recreating the StatefulSet with Non-cascade, downtime-free expansion was achieved.
A refusal circuit with a sparse gate-amplifier structure was repeatedly confirmed across 12 open-weight models.
ElevenLabs successfully scaled its AI voice marketplace by leveraging Stripe to solve complex payment and regulatory challenges.
Qwen3-Max-Thinking boosts reasoning performance through tool use and test-time scaling.