Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
Recent tweets highlight the emergence of Haiku, a new AI model comparable to GPT-3.5 but significantly cheaper and efficient. Haiku is derived from Opus and is seen as a strong competitor to GPT-4. The gap between small and large AI models is narrowing, with Haiku offering cost-effective inference and high performance.
Want to follow AI news, but just on new models and modeling benchmarks, take a look here. The model bleeds in some features and models, but pretty good I think * DBRX 132B for $10M * Claude Haiku rivals GPT4 and beats GPT3.5 * OpenAI testing monetization https://t.co/FsuuB9lX44 https://t.co/FRWUEl3B2q
Huge models that barely beat GPT 3.5 aren't as interesting as 7B models that can beat GPT 3.5 or Claude Haiku, which offers dirt-cheap inference and GPT-4 perf. Fine-tuning is also best done on smol models.
GPT-6 NEWS, Claude 3 Sentience, Meta's BIG Plans For AI, Super Realistic AI Video https://t.co/FSt27hpag3
