Update app.py

app.py

@@ -47,7 +47,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 # model_id = "mistralai/Mistral-7B-v0.3"
 
 model_id = "Qwen/Qwen1.5-14B-Chat"
-model_id = "Citaman/command-r-25-layer"
+# model_id = "Citaman/command-r-25-layer"
 
 
 tokenizer = AutoTokenizer.from_pretrained(
@@ -61,9 +61,9 @@ accelerator = Accelerator()
 model = AutoModelForCausalLM.from_pretrained(model_id, token= token, 
                                                  # torch_dtype= torch.uint8, 
                                              torch_dtype=torch.float16,
-                                              load_in_8bit=True,
+                                              # load_in_8bit=True,
                                              # #  # torch_dtype=torch.fl,
-                                             # attn_implementation="flash_attention_2",
+                                             attn_implementation="flash_attention_2",
                                              low_cpu_mem_usage=True,
                                              # device_map='cuda',
                                              # device_map=accelerator.device_map,