🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.
update readme
gongjy6e9cd28ef9b3
diff --git a/README.md b/README.mdindex 5b1c248..4b9dbf5 100644--- a/README.md+++ b/README.md@@ -770,7 +770,8 @@ minimind模型本身没有使用较大的数据集训练,也没有针对回答> 您的支持就是持续改进项目的动力> [!NOTE]-> 众擎易举。如果您已经尝试训练了新的MiniMind型号,欢迎在Discussions或Issues中分享您的模型权重<br/>+> 众人拾柴火焰高+> 如果您已经尝试训练了新的MiniMind型号,欢迎在Discussions或Issues中分享您的模型权重<br/>> 可以是在特定下游任务或垂直领域(例如情感识别、医疗、心理、金融、法律问答等)的MiniMind新模型版本<br/>> 也可以是拓展训练后(例如探索更长文本序列、更大体积(0.1B+)或更大的数据集)的MiniMind新模型版本<br/>> 任何分享都视作独一无二的,所有尝试都具有价值,并受到鼓励<br/>diff --git a/README_en.md b/README_en.mdindex f2e0aff..4cb7904 100644--- a/README_en.md+++ b/README_en.md@@ -849,16 +849,12 @@ your model with third-party UIs, such as fastgpt, OpenWebUI, etc.> [!NOTE]-> An individual's resources, energy, and time are limited, so we encourage everyone to participate and contribute-> collectively. If you have trained model weights, you are welcome to share them in the Discussions or Issues-> sections.<br/>-> These models can be new versions of MiniMind tailored for specific downstream tasks or vertical domains (such as-> sentiment recognition, healthcare, psychology, finance, legal Q&A, etc.).<br/>-> They can also be new versions of MiniMind models that have undergone extended training, exploring longer text-> sequences, larger volumes (such as 0.1B+), or more extensive datasets.<br/>-> Each contribution is unique, and all attempts are valuable and encouraged.<br/>-> Any shared contributions will be promptly recognized and compiled in the acknowledgments list. Thank you once again-> for everyone's support!+> Many hands make light work.+> If you have already tried training a new MiniMind model, you are welcome to share your model weights in Discussions or Issues. <br/>+> This can be a new version of MiniMind for a specific downstream task or vertical domain (e.g., sentiment recognition, healthcare, psychology, finance, legal Q&A, etc.). <br/>+> It can also be a new version of MiniMind after extended training (e.g., exploring longer text sequences, larger sizes (0.1B+), or larger datasets). <br/>+> Any contribution is considered unique and valuable, and all attempts are encouraged. <br/>+> These contributions will be promptly discovered and compiled in the acknowledgment list. Thank you again for all the support!## 🤝[Contributors](https://github.com/jingyaogong/minimind/graphs/contributors)