Update app.py

app.py

@@ -6,6 +6,7 @@ import gradio as gr
 import torch
 from utils import *
 from presets import *
+from transformers import GPT2Tokenizer, AutoModelForCausalLM
 
 #antwort=""
 ######################################################################
@@ -17,13 +18,22 @@ from presets import *
 
 #Alternativ mit beliebigen Modellen:
 #base_model = "project-baize/baize-v2-7b"  #load_8bit = False (in load_tokenizer_and_model)
-base_model = "alexkueck/li-tis-tuned-2"  #load_8bit = False (in load_tokenizer_and_model)
+#base_model = "alexkueck/li-tis-tuned-2"  #load_8bit = False (in load_tokenizer_and_model)
 #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 = "lmsys/vicuna-13b-v1.3"
-tokenizer,model,device = load_tokenizer_and_model(base_model,False)
+base_model = "gpt2-xl"
+
+####################################
+#Model und Tokenzier laden
+#tokenizer,model,device = load_tokenizer_and_model(base_model,False)
+
+################################
+#Alternativ: Model und Tokenizer direkt ladentokenizer = AutoTokenizer.from_pretrained(base_model, use_fast = True, use_auth_token=True)
+tokenizer = GPT2Tokenizer.from_pretrained(base_model, use_fast = True, use_auth_token=True)
+model = GPT2LMHeadModel.from_pretrained('gpt2')   # options: ['gpt2', 'gpt2-medium', 'gpt2-large', 'gpt2-xl']
 
 
 ########################################################################