fine tuning
fine tuning at Beyond Market Intelligence is a file of 27 stories. The newest of them: “Kimi K3 weights are open, but the license demands your attention”, “Engineering AI for mobile: balancing creativity, cost, and user trust”, and “Learning Rate Defaults Don't Fit Small Dataset Fine-Tuning”. Moonshot AI has released Kimi K3's full weights, and on the surface, it's a generous move for a 2.8-trillion-parameter model. Bhavuk Jain's session pulls back the curtain on what it takes to move AI from research to the phones in our pockets. Acme AI is the next-generation, AI-powered spreadsheet platform built to replace Excel and redefine how analysts, data scientists, and enterprise teams work… The list below is every fine tuning story on Beyond Market Intelligence, newest first.
Kimi K3 weights are open, but the license demands your attention
Moonshot AI has released Kimi K3's full weights, and on the surface, it's a generous move for a 2.8-trillion-parameter model. But the attached license is where the real story lives. It grants broad rights while quietly adding commercial strings for larger firms, think $20 million revenue thresholds and a "Model as a Service" definition that could snag unexpected players. Enterprises should read this carefully, because "open weights" no longer means "open for everyone." For most internal use, it's permissive.
Engineering AI for mobile: balancing creativity, cost, and user trust
Bhavuk Jain's session pulls back the curtain on what it takes to move AI from research to the phones in our pockets. He gets specific about the engineering behind AI Wallpapers and Circle to Search, tackling runtime guardrails and model latency without losing sight of the user. For leaders, it's a grounded look at balancing cost and experience.
Learning Rate Defaults Don't Fit Small Dataset Fine-Tuning
A single learning rate buried in a shared notebook cost someone three weeks of their life. The qlora 2e-4 default is wrong under 10k samples, and nobody talks about it. This user did the hard work of questioning the copy-paste consensus, and they found that dropping to 1e-4 and adding epochs moved the eval more than all their data cleaning combined. That is the practical insight most tutorials skip. If you are fighting small datasets, start lower and tune that one number.