diff --git a/pyproject.toml b/pyproject.toml
index 7e8956c712..cad8580449 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -34,7 +34,7 @@ exclude = ["images*"]
 [project.optional-dependencies]
 huggingface = [
     "transformers>=4.38.0",
-    "datasets",
+    "datasets>=2.16.0",
     "sentencepiece",
     "accelerate>=0.26.1",
     "trl>=0.7.9",