🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.
update images
gongjyf475e4e40721
diff --git a/README_en.md b/README_en.mdindex e349805..71bfc34 100644--- a/README_en.md+++ b/README_en.md@@ -1305,7 +1305,6 @@ scores:| 7 | E | 22 | 23 | 15 | 14 | 74 || 8 | G | 10 | 12 | 10 | 10 | 42 |----### 👉 Subjective Effect Summary@@ -1318,6 +1317,8 @@ My personal evaluation aligns with DeepSeek-R1's results,and:* Repeating the timeless Scaling Law: The larger the parameters and the more training data, the stronger the model'sperformance.+---+## Ⅲ Objective BenchmarkNow, onto the much-anticipated benchmark testing phase. We won’t bother comparing with models like Qwen or GLM-leveldiff --git a/images/minimind2.gif b/images/minimind2.gifindex bcf3eb5..43c9cd1 100644Binary files a/images/minimind2.gif and b/images/minimind2.gif differ