Data Prep: Successfully converted training data to JSONL format for AutoTrain

convert_to_jsonl.py

@@ -0,0 +1,40 @@
+import json
+import os
+
+input_path = 'src/environment/training_samples.jsonl'
+output_path = 'src/environment/training_samples_final.jsonl'
+
+print(f"Reading {input_path}...")
+with open(input_path, 'r', encoding='utf-8') as f:
+    content = f.read().strip()
+
+# Restore brackets if missing
+if not content.startswith('['):
+    content = '[' + content
+if not content.endswith(']'):
+    content = content + ']'
+
+try:
+    data = json.loads(content)
+except Exception as e:
+    # If the trailing comma exists, try to remove it
+    if content.endswith(',]'):
+        content = content[:-2] + ']'
+        data = json.loads(content)
+    else:
+        raise e
+
+print(f"Converting {len(data)} entries...")
+with open(output_path, 'w', encoding='utf-8') as f:
+    for entry in data:
+        text_content = (
+            f"### State\n"
+            f"Confusion: {entry.get('confusion')}\n"
+            f"Action: {entry.get('action')}\n"
+            f"Reward: {entry.get('reward')}\n"
+            f"Next Confusion: {entry.get('next_confusion')}"
+        )
+        json.dump({"text": text_content}, f, ensure_ascii=False)
+        f.write('\n')
+
+print(f"Done! Successfully created {output_path}")