A research paper on LoRA (Low-Rank Adaptation) concluded that the technique exhibits a trade-off where it "learns less but forgets less," meaning it is less prone to catastrophic forgetting but also a...

The open-source model GPT-OSS is considered the first open-weight model trained specifically with tool use in mind.

Open-weight models from China are popular partly because they have unrestricted open licenses, unlike models like Llama or Gemma which have usage restrictions for large companies.

The research for the "recursive language model" paper was conducted entirely using GPT-4, without the use of local models, due to the high compute budget required.

The '996' work culture, defined as working from 9 a.m. to 9 p.m. six days a week, is becoming a trend in Silicon Valley AI companies.

In response to U.S. sanctions, Chinese telecommunications company Huawei has successfully developed its own advanced semiconductor chips.

The Qwen3 Next model added a gated deltanet, inspired by state-space models, to make its attention mechanism scale linearly with inference token prediction.