Update app.py

app.py

@@ -22,7 +22,7 @@ MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
 HF_TOKEN = os.getenv("HF_TOKEN")
 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
 
-model_name = "Qwen/Qwen2.5-3B-Instruct"
+model_name = "ehristoforu/cq3bitfp32"
 
 model = AutoModelForCausalLM.from_pretrained(
     model_name,
@@ -30,9 +30,10 @@ model = AutoModelForCausalLM.from_pretrained(
 )
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 
-peft_model = AutoPeftModelForCausalLM.from_pretrained("ehristoforu/CoolQwen2.5-3b-it")
-merged_model = peft_model.merge_and_unload()
-merged_model.save_pretrained("./coolqwen")
+#peft_model = AutoPeftModelForCausalLM.from_pretrained("ehristoforu/CoolQwen2.5-3b-it")
+#merged_model = peft_model.merge_and_unload()
+#merged_model.save_pretrained("./coolqwen")
+model.save_pretrained("./coolqwen")
 tokenizer.save_pretrained("./coolqwen")
 
 from huggingface_hub import HfApi
@@ -43,7 +44,7 @@ api = HfApi()
 
 api.upload_folder(
     folder_path="./coolqwen",
-    repo_id="ehristoforu/CoolQwen2.5-3B-IT-fp16",
+    repo_id="ehristoforu/coolqwen-3b-it",
     repo_type="model",
     token=HF_TOKEN,
 )