feat: Use IterableDataset for true streaming (no disk writes)

long-context-pdfs.py

@@ -1,7 +1,7 @@
 # /// script
 # requires-python = ">=3.10"
 # dependencies = [
-#     "polars",
+#     "polars>=1.0",
 #     "datasets",
 # ]
 # ///
@@ -30,7 +30,19 @@ Examples:
 
 import argparse
 import polars as pl
-from datasets import Dataset
+from datasets import Features, IterableDataset, Value
+
+FEATURES = Features(
+    {
+        "id": Value("string"),
+        "url": Value("string"),
+        "text": Value("string"),
+        "language": Value("string"),
+        "token_count": Value("int64"),
+        "dump": Value("string"),
+        "page_average_lid_score": Value("float64"),
+    }
+)
 
 
 def polars_to_generator(lf: pl.LazyFrame, chunk_size: int = 100):
@@ -157,12 +169,12 @@ def main():
         lf = lf.limit(args.limit)
 
     print(f"\nStreaming to dataset (chunk_size={args.chunk_size})...")
-    ds = Dataset.from_generator(
-        lambda: polars_to_generator(lf, chunk_size=args.chunk_size)
+    ds = IterableDataset.from_generator(
+        lambda: polars_to_generator(lf, chunk_size=args.chunk_size),
+        features=FEATURES,
     )
-    print(f"Created dataset with {len(ds)} rows")
 
-    print(f"\nPushing to {args.output}...")
+    print(f"\nPushing to {args.output} (streaming)...")
     ds.push_to_hub(args.output, private=args.private)
     print(f"\nDone! https://huggingface.co/datasets/{args.output}")