ZhouChuYue commited on
Commit
6f63dcd
·
1 Parent(s): 415139a

Add pp (percentage point) unit to improvement scores in README

Browse files
Files changed (2) hide show
  1. README.md +1 -1
  2. README_ZH.md +1 -1
README.md CHANGED
@@ -60,7 +60,7 @@ To address these issues, we propose ***UltraData-Math***—a large-scale high-qu
60
  - **L2 Selected Data**: Uses proprietary large models to annotate seed data and distills it into a lightweight embedding classifier to achieve efficient quality grading of the full corpus.
61
  - **L3 Refined Data**: Produces structured content with clear reasoning through rewriting, synthetic generation, and refinement in various formats such as Q&A, multi-turn dialogues, multi-style rewriting, and knowledge-grounded textbooks.
62
 
63
- Experiments show that on the MiniCPM-1.2B architecture, ***UltraData-Math*** achieves a score of **37.02** on the MATH500 benchmark, an improvement of **+3.62** compared to Nemotron-CC 4plus; it achieves **61.79** on GSM8K, an improvement of **+3.34**, while maintaining code generation and general knowledge capabilities.
64
 
65
  ***UltraData-Math*** has been applied to the mathematical pre-training of the [MiniCPM Series](https://huggingface.co/collections/openbmb/minicpm-4-6841ab29d180257e940baa9b) models.
66
 
 
60
  - **L2 Selected Data**: Uses proprietary large models to annotate seed data and distills it into a lightweight embedding classifier to achieve efficient quality grading of the full corpus.
61
  - **L3 Refined Data**: Produces structured content with clear reasoning through rewriting, synthetic generation, and refinement in various formats such as Q&A, multi-turn dialogues, multi-style rewriting, and knowledge-grounded textbooks.
62
 
63
+ Experiments show that on the MiniCPM-1.2B architecture, ***UltraData-Math*** achieves a score of **37.02** on the MATH500 benchmark, an improvement of **+3.62 pp** compared to Nemotron-CC 4plus; it achieves **61.79** on GSM8K, an improvement of **+3.34 pp**, while maintaining code generation and general knowledge capabilities.
64
 
65
  ***UltraData-Math*** has been applied to the mathematical pre-training of the [MiniCPM Series](https://huggingface.co/collections/openbmb/minicpm-4-6841ab29d180257e940baa9b) models.
66
 
README_ZH.md CHANGED
@@ -25,7 +25,7 @@
25
  - **L2 精选数据层**:使用闭源大模型标注种子数据并蒸馏至轻量 embedding 分类器,实现全量语料的高效质量分级。
26
  - **L3 精炼数据层**:通过改写、合成生成与精炼,生成具有清晰推理链条的结构化内容,涵盖 Q&A、多轮对话、多风格改写、知识教材等多种格式。
27
 
28
- 实验表明,在 MiniCPM-1.2B 架构上,***UltraData-Math*** 在 MATH500 基准上达到 **37.02** 分,相较 Nemotron-CC 4plus 提升 **+3.62** 分;在 GSM8K 上达到 **61.79** 分,提升 **+3.34** 分,同时保持代码生成与通用知识能力。
29
 
30
  ***UltraData-Math*** 已应用于 [MiniCPM 系列](https://huggingface.co/collections/openbmb/minicpm-4-6841ab29d180257e940baa9b) 模型的数学预训练。
31
 
 
25
  - **L2 精选数据层**:使用闭源大模型标注种子数据并蒸馏至轻量 embedding 分类器,实现全量语料的高效质量分级。
26
  - **L3 精炼数据层**:通过改写、合成生成与精炼,生成具有清晰推理链条的结构化内容,涵盖 Q&A、多轮对话、多风格改写、知识教材等多种格式。
27
 
28
+ 实验表明,在 MiniCPM-1.2B 架构上,***UltraData-Math*** 在 MATH500 基准上达到 **37.02** 分,相较 Nemotron-CC 4plus 提升 **+3.62 pp**;在 GSM8K 上达到 **61.79** 分,提升 **+3.34 pp**,同时保持代码生成与通用知识能力。
29
 
30
  ***UltraData-Math*** 已应用于 [MiniCPM 系列](https://huggingface.co/collections/openbmb/minicpm-4-6841ab29d180257e940baa9b) 模型的数学预训练。
31