Update README.md

README.md

@@ -21,4 +21,39 @@ configs:
     path: data/train-*
   - split: test
     path: data/test-*
+language:
+- ca
 ---
+
+## Dataset Card for GSM8K (Catalan Version)
+
+### Dataset Summary
+GSM8K (Grade School Math 8K) is a dataset containing 8.5K high-quality math word problems designed to assess multi-step mathematical reasoning in language models. This version is a Catalan translation of the original `openai/gsm8k` dataset, preserving the structure and objectives of the original English dataset.
+
+Key features of the dataset:
+- Problems require between 2 and 8 steps to solve.
+- Solutions involve sequences of elementary calculations using basic arithmetic operations (+, −, ×, ÷) to reach the final answer.
+- A bright middle school student should be able to solve every problem: according to the original paper, "Problems require no concepts beyond the level of early Algebra, and the vast majority of problems can be solved without explicitly defining a variable."
+- Solutions are provided in natural language rather than purely mathematical expressions.
+
+### Supported Tasks and Leaderboards
+This dataset is mainly used to test logic and mathematical reasoning in language modeling. It has been used in multiple benchmarks, including the LLM Leaderboard.
+
+### Languages
+The dataset text is in Catalan, whereas the original dataset was in English (`BCP-47: en`).
+
+### Dataset Structure
+#### Data Instances
+For the main configuration, each instance contains a grade-school level math question and a detailed answer with multiple reasoning steps and calculator annotations.
+
+```json
+{
+    "question": "Natalia vendió clips a 48 de sus amigos en abril, y luego vendió la mitad de esa cantidad en mayo. ¿Cuántos clips vendió Natalia en total entre abril y mayo?",
+    "answer": "Natalia vendió 48/2 = <<48/2=24>>24 clips en mayo.\nNatalia vendió 48+24 = <<48+24=72>>72 clips en total entre abril y mayo.\n#### 72"
+}
+```
+
+### Data Fields
+The data fields remain consistent across different configurations:
+- **question**: A grade school math problem presented as a natural language question.
+- **answer**: The full solution string to the question, containing multiple reasoning steps, calculator annotations, and the final numeric solution.