Update app.py

app.py

@@ -24,22 +24,22 @@ from transformers import LlamaForCausalLM, LlamaTokenizer
 #base_model = "TheBloke/airoboros-13B-HF"  #load_8bit = False (in load_tokenizer_and_model)
 #base_model = "EleutherAI/gpt-neo-1.3B"    #load_8bit = False (in load_tokenizer_and_model)
 #base_model = "TheBloke/airoboros-13B-HF"   #load_8bit = True
-#base_model = "TheBloke/vicuna-13B-1.1-HF"   #load_8bit = ?
+base_model = "TheBloke/vicuna-13B-1.1-HF"   #load_8bit = ?
 #following runs only on GPU upgrade
-base_model = "TheBloke/airoboros-65B-gpt4-1.3-GPTQ"  #model_basename = "airoboros-65b-gpt4-1.3-GPTQ-4bit--1g.act.order"
+#base_model = "TheBloke/airoboros-65B-gpt4-1.3-GPTQ"  #model_basename = "airoboros-65b-gpt4-1.3-GPTQ-4bit--1g.act.order"
 #base_model = "lmsys/vicuna-13b-v1.3"
 #base_model = "gpt2-xl"   # options: ['gpt2', 'gpt2-medium', 'gpt2-large', 'gpt2-xl']
 
 ####################################
 #Model und Tokenzier laden
-#tokenizer,model,device = load_tokenizer_and_model(base_model,False)
+tokenizer,model,device = load_tokenizer_and_model(base_model,False)
 
 ################################
 #Alternativ: Model und Tokenizer für GPT2
 #tokenizer,model,device = load_tokenizer_and_model_gpt2(base_model,False)
 
 #Alternativ bloke gpt3 und4 - only with GPU upgarde
-tokenizer,model,device = load_tokenizer_and_model_bloke_gpt(base_model, "airoboros-65b-gpt4-1.3-GPTQ-4bit--1g.act.order")
+#tokenizer,model,device = load_tokenizer_and_model_bloke_gpt(base_model, "airoboros-65b-gpt4-1.3-GPTQ-4bit--1g.act.order")
   
 #Alternativ Model und Tokenzier laden für Baize
 #tokenizer,model,device = load_tokenizer_and_model_Baize(base_model,False)