Skip to content
Python

Update section headers in README.md

QI CHENd5f3a769ee29
diff --git a/README.md b/README.md
index f98daef..88fbddc 100644
--- a/README.md
+++ b/README.md
@@ -43,12 +43,12 @@ The pipeline of our model is shown in Fig. 1 which consists of three stages:
## 🌈Data Preprocessing
We construct the CBT Cognitive Triplet Dataset based on [PsyQA](https://github.com/thu-coai/PsyQA)
-## Stage 1: Cognitive Distortion Identification
+# Stage 1: Cognitive Distortion Identification
From the original PsyQA dataset (~22K samples), we selected 9,437 question samples exhibiting cognitive distortions through a carefully designed prompt-based filtering process.
Model: DeepSeek
Prompt: [filter_prompt.txt](prompts/filter_prompt.txt) (Prompt1 Version 2)​
Script: [identify_distortion.py](src/identify_distortion.py)
-## Stage 2: Rational Response Generation
+# Stage 2: Rational Response Generation
We applied a structured prompt-based generation procedure to construct the CBT Cognitive Triplet Dataset, resulting in 9,437 validated samples.
Model: ChatGPT (GPT-5 mini)
Prompt: [response_prompt.txt](prompts/response_prompt.txt) (Prompt2 Version 3)