Jackrong commited on
Commit
1e03e15
·
verified ·
1 Parent(s): 6867bad

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -0
README.md CHANGED
@@ -595,6 +595,11 @@ Through analysis of the model's actual outputs, we can see that it has "evolved"
595
 
596
  - ✅ Through behaviors like deductive refutation, strategy evaluation, and self-correction, the model has proven that its reasoning ability is no longer mere format imitation but has evolved into a genuine, self-driven logical thinking capability.
597
 
 
 
 
 
 
598
 
599
  ### Limitations
600
 
@@ -1230,6 +1235,11 @@ Question: Who is the doctor?
1230
  - 由于我的训练数据涵盖了中英双语数据集,且基座模型本身也经过多语言训练,模型在推理过程中可能存在语言混合的问题,即在同一个思维链或答案中同时使用中文和英文。或者推理完成后答案输出不完整。 尽管我在后期训练中尝试缓解此问题,但这仍是一个需要进一步改进的方面。
1231
  - 尽管模型在代数应用题等领域表现良好,但其推理能力可能并不均衡。例如,在一些更专业的领域,或者通用聊天等方面(后续的SFT方面提升有限),其能力可能相对较弱。此外,当前模型不具备调用外部工具的能力,这限制了其在解决需要精确计算或获取实时外部知识的复杂问题时的上限。
1232
 
 
 
 
 
 
1233
 
1234
  ### Prompt Template Format
1235
 
 
595
 
596
  - ✅ Through behaviors like deductive refutation, strategy evaluation, and self-correction, the model has proven that its reasoning ability is no longer mere format imitation but has evolved into a genuine, self-driven logical thinking capability.
597
 
598
+ ### How to use
599
+
600
+ * **‼️Add training system prompt for reasoning**
601
+ - 'system prompt' : You are Phi, a language model trained by Microsoft to help users. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format: <think> Thought section </think> Solution section. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion. Now, try to solve the following question through the above guidelines.
602
+
603
 
604
  ### Limitations
605
 
 
1235
  - 由于我的训练数据涵盖了中英双语数据集,且基座模型本身也经过多语言训练,模型在推理过程中可能存在语言混合的问题,即在同一个思维链或答案中同时使用中文和英文。或者推理完成后答案输出不完整。 尽管我在后期训练中尝试缓解此问题,但这仍是一个需要进一步改进的方面。
1236
  - 尽管模型在代数应用题等领域表现良好,但其推理能力可能并不均衡。例如,在一些更专业的领域,或者通用聊天等方面(后续的SFT方面提升有限),其能力可能相对较弱。此外,当前模型不具备调用外部工具的能力,这限制了其在解决需要精确计算或获取实时外部知识的复杂问题时的上限。
1237
 
1238
+ ### ‼️添加训练时的系统提示词激活推理
1239
+ - 'system prompt' : You are Phi, a language model trained by Microsoft to help users. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format: <think> Thought section </think> Solution section. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion. Now, try to solve the following question through the above guidelines.
1240
+
1241
+
1242
+
1243
 
1244
  ### Prompt Template Format
1245