update readme

README.md

@@ -19,3 +19,13 @@ configs:
   - split: test
     path: data/test-*
 ---
+
+# MNLP Quantization Dataset
+
+This dataset is a copy of the MCQA dataset used during our project and reused here to calibrate our quantized model using post-training quantization (PTQ).
+
+## Origin
+Originally hosted at: [`hssawhney/MNLP-Dataset`](https://huggingface.co/datasets/hssawhney/MNLP-Dataset)
+
+## Use
+Used to calibrate our quantized W8A8 MCQA model (see report, Section 3.2). The quantization setup used 512 examples from the test set.