diff --git a/data/openwebtext/prepare.py b/data/openwebtext/prepare.py
index 2a9b9752b7..f7ed164ecc 100644
--- a/data/openwebtext/prepare.py
+++ b/data/openwebtext/prepare.py
@@ -1,6 +1,7 @@
 # saves the openwebtext dataset to a binary file for training. following was helpful:
 # https://github.com/HazyResearch/flash-attention/blob/main/training/src/datamodules/language_modeling_hf.py
 
+import multiprocessing
 import os
 from tqdm import tqdm
 import numpy as np
@@ -9,7 +10,7 @@
 
 # number of workers in .map() call
 # good number to use is ~order number of cpu cores // 2
-num_proc = 8
+num_proc = multiprocessing.cpu_count() // 2
 
 # number of workers in load_dataset() call
 # best number might be different from num_proc above as it also depends on NW speed.