Update app.py

app.py

@@ -29,17 +29,17 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 peft_model_id = "charansr/llama2-7b-chat-hf-therapist"
 
 config = PeftConfig.from_pretrained(peft_model_id,
-                                   use_auth_token="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", load_in_8bit=True, device_map='cpu',)
+                                   use_auth_token="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", load_in_8bit=True).to("cpu")
 
-newmodel = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, return_dict=True, load_in_8bit=True, device_map='cpu',
-                                                use_auth_token="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")
+newmodel = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, return_dict=True, load_in_8bit=True,
+                                                use_auth_token="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").to("cpu")
 
 newtokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path,
-                                            use_auth_token="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", load_in_8bit=True, device_map='cpu',)
+                                            use_auth_token="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", load_in_8bit=True).to("cpu")
 
 # Load the Lora model
 newmodel = PeftModel.from_pretrained(newmodel, peft_model_id,
-                                    use_auth_token="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", load_in_8bit=True, device_map='cpu')
+                                    use_auth_token="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", load_in_8bit=True).to("cpu")
 
 def givetext(input_text,lmodel,ltokenizer):
     try:
@@ -61,13 +61,13 @@ def givetext(input_text,lmodel,ltokenizer):
         with torch.no_grad():
             #print("BEFORE RETURNING")
 
-            print("BEFORE ATTEMPTING TO MOVE LMODEL TO CPU")
+            #print("BEFORE ATTEMPTING TO MOVE LMODEL TO CPU")
             
-            lmodel = lmodel.to("cpu")
+            #lmodel = lmodel.to("cpu")
 
-            print("BEFORE ATTEMPTING .cpu()")
+            #print("BEFORE ATTEMPTING .cpu()")
 
-            lmodel.cpu()
+            #lmodel.cpu()
             
             print("BEFORE GENERATING LMODEL")