Upload code/build_sft_v20_dataset.py with huggingface_hub

code/build_sft_v20_dataset.py

@@ -609,7 +609,7 @@ def build_domain_reasoning():
     if len(pool) < cap:
         print(f"  Loading open-thoughts/OpenThoughts (need {cap - len(pool)} more)...")
         try:
-            ds_ot = load_dataset("open-thoughts/OpenThoughts-114k", split="train", streaming=True)
+            ds_ot = load_dataset("open-thoughts/Open-Thoughts-114k", split="train", streaming=True)
             CODE_KWS = {"python", "code", "function", "sql", "algorithm", "class ", "def "}
             MATH_KWS = {"calculate", "solve", "equation", "integer", "prime", "probability"}
             for row in ds_ot: