""I don't even agree with the assumption that Chinese models are catching up. I believe we're still far behind. I guess the gap is getting larger, very sadly.""
""On paper, every big tech company in China has a good model. But I don't think they're good enough.""
""Google can train or perform a full round of LLM training, both pre-training and post-training, in three months. But ByteDance probably we can only do one iteration in half a year.""
""There's a huge difference between the infrastructure at Google and ByteDance. I don't think we're getting high-quality data.""
Zhang Chi, a former ByteDance engineer, asserts that China is not catching up to the US in AI but is actually falling further behind. He emphasizes that while Chinese companies may perform well on benchmarks, their models do not translate effectively to real-world applications. Zhang highlights issues such as slower iteration speeds, with US companies like Google able to train models much faster. He also points out structural disadvantages in China, including access to advanced chips and lower-quality training data.
Read at www.businessinsider.com
Unable to calculate read time
Collection
[
|
...
]