Initial commit

app.py

@@ -20,15 +20,15 @@ if not hf_token:
 login(token=hf_token)
 logger.debug("Hugging Face login successful")
 
-# لود مدل CodeLlama
-logger.debug("Loading meta-llama/CodeLlama-7b-Instruct-hf model")
+# لود مدل Gemma 3 4B
+logger.debug("Loading google/gemma-3-4b-it model")
 try:
-    tokenizer = AutoTokenizer.from_pretrained("meta-llama/CodeLlama-7b-Instruct-hf", cache_dir="/tmp/hf_cache")
-    model = AutoModelForCausalLM.from_pretrained("meta-llama/CodeLlama-7b-Instruct-hf", cache_dir="/tmp/hf_cache", torch_dtype=torch.float16, device_map="auto")
+    tokenizer = AutoTokenizer.from_pretrained("google/gemma-3-4b-it", cache_dir="/tmp/hf_cache")
+    model = AutoModelForCausalLM.from_pretrained("google/gemma-3-4b-it", cache_dir="/tmp/hf_cache", torch_dtype=torch.bfloat16, device_map="auto")
     code_gen = pipeline("text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512, temperature=0.6, top_p=0.95)
-    logger.debug("meta-llama/CodeLlama-7b-Instruct-hf model loaded")
+    logger.debug("google/gemma-3-4b-it model loaded")
 except Exception as e:
-    logger.error(f"Failed to load CodeLlama model: {e}")
+    logger.error(f"Failed to load Gemma 3 model: {e}")
     raise
 
 # آزادسازی حافظه