News Microsoft reveals Phi-3.5 — this new small AI model outperforms Gemini and GPT-4o

[admin]

Microsoft's new Phi-3.5 is a powerful small language model capable of math and reasoning performance equal to models like GPT-4o mini or Gemini Flash 1.5.

Microsoft reveals Phi-3.5 — this new small AI model outperforms Gemini and GPT-4o : Read more

 

[Araki]

Nothing's exciting about it. It is censored beyond repair, even finetuning can't fix it. The previous iteration straight-up did not contain any unsafe content in the dataset, consisting of toddler-friendly AI-generated textbooks. For this iteration, I'm certain they went even further and yet again wasted Earth's energy resources on training another set of infantile LLMs.