🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.
[update] image
jingyaogong88e675dc2c20
diff --git a/images/benchmark_radar.jpg b/images/benchmark_radar.jpgindex a8b231b..3d1d664 100644Binary files a/images/benchmark_radar.jpg and b/images/benchmark_radar.jpg differ