change model

app.py

@@ -10,6 +10,7 @@ logger = logging.get_logger("transformers")
 model_name = "openai-community/gpt2" #"openai-community/gpt2" or "TheBloke/Mistral-7B-Instruct-v0.1-GPTQ" or "TheBloke/Llama-2-7B-Chat-GGML" or "TheBloke/zephyr-7B-beta-GPTQ"
 tokenizer = AutoTokenizer.from_pretrained(model_name)#,use_fast=True
 model = AutoModelForCausalLM.from_pretrained(model_name)#,device_map="auto",trust_remote_code=False,revision="main")
+tokenizer.pad_token_id = tokenizer.eos_token_id
 
 #transfer model on GPU
 #model.to("cuda")
@@ -18,8 +19,9 @@ model = AutoModelForCausalLM.from_pretrained(model_name)#,device_map="auto",trus
 #@spaces.GPU(duration=60)
 def generate_text(input_text):
     input_ids = tokenizer.encode(input_text, return_tensors="pt")#.to("cuda")
-    #attention_mask = input_ids.ne(tokenizer.pad_token_id).long()
-    output = model.generate(input_ids, max_new_tokens=512, top_k=50, top_p=0.95, temperature=0.7, do_sample=True)# attention_mask=attention_mask, max_length=100, num_return_sequences=1, no_repeat_ngram_size=2, top_k=50, top_p=0.95, temperature=0.7, do_sample=True)
+    attention_mask = input_ids.ne(tokenizer.pad_token_id).long()
+    #output = model.generate(input_ids, max_new_tokens=512, top_k=50, top_p=0.95, temperature=0.7, do_sample=True)# attention_mask=attention_mask, max_length=100, num_return_sequences=1, no_repeat_ngram_size=2, top_k=50, top_p=0.95, temperature=0.7, do_sample=True)
+    output = model.generate(input_ids, attention_mask=attention_mask, max_length=100, num_return_sequences=1, no_repeat_ngram_size=2, top_k=50, top_p=0.95, temperature=0.7, do_sample=True)
     return tokenizer.decode(output[0])
 
 # Example of disabling Exllama backend (if applicable in your configuration)