mod dataset

2025-02-25 23:06:17 +01:00 · 2025-02-25 23:06:17 +01:00 · 5f06f859a5
parent 1f24b7467b
commit 5f06f859a5
1 changed files with 3 additions and 1 deletions
--- a/hft.py
+++ b/hft.py
@ -207,7 +207,9 @@ def main():
        print("\nBrak danych do treningu!")
        return

-    dataset = Dataset.from_list(data)
+    #dataset = Dataset.from_list(data)
+    dataset = Dataset.from_dict({k: [d[k] for d in data] for k in data[0]})
+

    def tokenize_function(examples):
        tokenized = tokenizer(