Update app.py

app.py

@@ -37,7 +37,8 @@ def generate(model,tokenizer,instruction,context):
             max_new_tokens=max_length,
             temperature=temperature,
             top_k=10,
-            repetition_penalty=repetition_penalty
+            repetition_penalty=repetition_penalty,
+            eos_token_id=tokenizer.eos_token_id
         )
     outputs = []
     for o in generation_output.sequences:
@@ -50,15 +51,15 @@ st.write(" ")
 
 instruction = st.text_area('Instrucción')
 context = st.text_area('Contexto')
-max_length = st.number_input('Max generation length')
+max_length = st.number_input('Max generation length',min=10,value=52)
 
 
+if (model == None):
+    model = AutoModelForCausalLM.from_pretrained("lagy/carballo_finetuned")
+    tokenizer = AutoTokenizer.from_pretrained("lagy/carballo_finetuned")
+    model.eval()
 
-model = AutoModelForCausalLM.from_pretrained("lagy/carballo_finetuned")
-tokenizer = AutoTokenizer.from_pretrained("lagy/carballo_finetuned")
-model.eval()
-
-if st.button('Generate') and max_length and instruction:
+if st.button('Generate'):
     #st.json(out)
     st.write("Generating...")
     output = generate(model,tokenizer,instruction,context)