updated model.py

model.py

@@ -1,5 +1,5 @@
 # model.py - Optimized version
-from transformers import AutoTokenizer, AutoModelForCausalLM
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 import torch
 from functools import lru_cache
 import os
@@ -60,7 +60,7 @@ def load_model_sync():
         
         # Load model with optimizations
         logger.info("🧠 Loading model...")
-        _model = AutoModelForCausalLM.from_pretrained(
+        _model = AutoModelForSeq2SeqLM.from_pretrained(
             model_id,
             trust_remote_code=config["trust_remote_code"],
             torch_dtype=config["torch_dtype"],