Update README.md

README.md

@@ -43,3 +43,18 @@ configs:
   - split: test
     path: data/test-*
 ---
+# QQP (Modified for LLaMA 1B)
+
+This dataset is a modified version of the **Quora Question Pairs (QQP)** dataset for duplicate question classification.
+
+## **Modifications:**
+- Labels were originally `"duplicate"` and `"not duplicate"`, now converted to integers (`0` and `1`).
+- The `question1` and `question2` columns were renamed to `text1` and `text2`.
+- Each example includes a task prefix: `Task: QQP | Q1: ... Q2: ...`
+- Tokenized using the **LLaMA-1B tokenizer**.
+- Maximum sequence length is **128 tokens**.
+
+## **Dataset Usage:**
+```python
+from datasets import load_dataset
+dataset = load_dataset("emirhanboge/qqp_llama1b_modified")