Update main.py

main.py

@@ -38,7 +38,7 @@ from pydantic import BaseModel, Field
 # ─────────────────────────────────────────────────────────────────────────
 
 BASE_DIR      = Path(__file__).parent
-INFERENCE_EXE = BASE_DIR / "inference.exe"
+INFERENCE_EXE = BASE_DIR / "inference"
 MODEL_BIN     = BASE_DIR / "model.bin"
 
 # FIX #7 — Chat template MUST match your SFT training data format exactly.