精选70Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)模型发布/更新
GLM称OPD训练成本可忽略不计
we completely missed this. Might be the biggest news from GLM. They do say "the…
we completely missed this. Might be the biggest news from GLM. They do say "the entire OPD training". Uh. @jietang could you… please… share more details? This basically means that OPD is negligibly cheap relative to pretraining and expert RL, right? And expert RL is parallel…
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力