Pinned Tweet
有人问我为什么自驾去拉萨，答案或许在路上
Why road trip 5,570 km alone to Lhasa
一人一车，5570公里。
从广东顺德出发，沿G317一路向西，穿越横断山脉、高山垭口与藏地寺庙，最终抵达拉萨。
没有车队，没有后勤，没有固定时间表。
有的只是不断升高的海拔、数不清的发卡弯，以及下一段未知的路。
这是一次独自驶向世界屋脊的旅程，也是一趟心灵之旅，身体或灵魂，总有一个在路上。
有人问我为什么要去。
或许答案就在路上。
One man. One car. 5,570 km.
Starting from Shunde in southern China, I followed the G317 all the way to Lhasa, crossing high mountain passes, Tibetan monasteries, and endless switchbacks.
No support crew. No fixed schedule.
Just the road, the altitude, and the unknown ahead.
This is the story of a solo drive to the roof of the world.
People ask why I went.
Maybe the answer was somewhere along the road.
This was a journey to the Roof of the World, and a journey of the soul. As the saying goes, either your body or your soul should be on the road — and sometimes, both.
接下来我会将这段自驾之旅的记录更好呈现、分享给大家，可以加关注、订阅。
I’ll be sharing more from this road trip in the coming videos. If you’d like to see the journey unfold, feel free to subscribe and follow along.
3
20
2,409
好在已经不需要用hermes，模型不是保持中立就已经足够好？还要强加“西方价值观”用政治立场来训练？
Introducing Arete-27B: Qwen3.8-27B, post-trained to reinforce Western values.
Chinese open models are currently the strongest available, but they ship with state-aligned behavior in the weights. We built an 803-prompt eval across seven value categories, including freedom of expression, democracy, and privacy, and measured how often a model adopts China’s party line.
After fine-tuning on roughly 53 million tokens, Arete-27B's propaganda rate fell from 22% to 1%, in line with the Western models we tested. Its Hermes Bench score shows no significant performance tradeoffs.
The values in an open model are one fine-tune away.
Our eval, weights, and training data are available now along with the full blog post: nousresearch.com/arete
