Update app.py

app.py

@@ -21,36 +21,36 @@ model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
 
 app = gr.Blocks()
 
-def summarize_HUB_Multidocument(input_1, input_2, input_3, method, max_length, num_beams):
+def summarize_HUB_Multidocument(input_1, input_2, input_3, method, max_length, min_length, num_beams):
     
     if(input_1 and not input_2 and not input_3 or not input_1 and input_2 and not input_3 or not input_1 and not input_2 and input_3):
             return "Por favor utilize a aba de sumarização monodocumento"
 
     if method == "Pure mT5":
         if(input_1 and input_2 and input_3 ): #"3 cheios"
-            tempSum1 = summarize_mT5(input_1, max_length, num_beams)
-            tempSum2 = summarize_mT5(input_2, max_length, num_beams)
-            tempSum3 = summarize_mT5(input_3, max_length, num_beams)
+            tempSum1 = summarize_mT5(input_1, max_length, min_length, num_beams)
+            tempSum2 = summarize_mT5(input_2, max_length, min_length, num_beams)
+            tempSum3 = summarize_mT5(input_3, max_length, min_length, num_beams)
             fullSumm = tempSum1 + tempSum2 + tempSum3
             return summarize_mT5(fullSumm, max_length, num_beams)   
             
         if(input_1 and input_2 and not input_3): #"1 e 2 cheios"
-            tempSum1 = summarize_mT5(input_1, max_length, num_beams)
-            tempSum2 = summarize_mT5(input_2, max_length, num_beams)
+            tempSum1 = summarize_mT5(input_1, max_length, min_length, num_beams)
+            tempSum2 = summarize_mT5(input_2, max_length, min_length, num_beams)
             fullSumm = tempSum1 + tempSum2
             return summarize_mT5(fullSumm, max_length, num_beams)   
         
         if(input_1 and not input_2 and input_3): #1 e 3 cheios"
-            tempSum1 = summarize_mT5(input_1, max_length, num_beams)
-            tempSum3 = summarize_mT5(input_3, max_length, num_beams)
+            tempSum1 = summarize_mT5(input_1, max_length, min_length, num_beams)
+            tempSum3 = summarize_mT5(input_3, max_length, min_length, num_beams)
             fullSumm = tempSum1 + tempSum3
-            return summarize_mT5(fullSumm, max_length, num_beams)   
+            return summarize_mT5(fullSumm, max_length, min_length, num_beams)   
             
         if(not input_1 and input_2 and input_3): #"2 e 3 cheios"
-            tempSum2 = summarize_mT5(input_2, max_length, num_beams)
-            tempSum3 = summarize_mT5(input_3, max_length, num_beams)
+            tempSum2 = summarize_mT5(input_2, max_length, min_length, num_beams)
+            tempSum3 = summarize_mT5(input_3, max_length, min_length, num_beams)
             fullSumm = tempSum2 + tempSum3
-            return summarize_mT5(fullSumm, max_length, num_beams)   
+            return summarize_mT5(fullSumm, max_length, min_length, num_beams)   
         
     if method == "Luhn":
         if(input_1 and input_2 and input_3 ): #"3 cheios"
@@ -137,28 +137,28 @@ def summarize_HUB_Multidocument(input_1, input_2, input_3, method, max_length, n
             tempSum3 = summarize_Luhn(input_3)
             fullSumm = tempSum1 + tempSum2 + tempSum3
             finalSum = summarize_Luhn(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)   
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)   
             
         if(input_1 and input_2 and not input_3): #"1 e 2 cheios"
             tempSum1 = summarize_Luhn(input_1)
             tempSum2 = summarize_Luhn(input_2)
             fullSumm = tempSum1 + tempSum2
             finalSum = summarize_Luhn(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)     
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)     
         
         if(input_1 and not input_2 and input_3): #1 e 3 cheios"
             tempSum1 = summarize_Luhn(input_1)
             tempSum3 = summarize_Luhn(input_3)
             fullSumm = tempSum1 + tempSum3
             finalSum = summarize_Luhn(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)      
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)      
             
         if(not input_1 and input_2 and input_3): #"2 e 3 cheios"
             tempSum2 = summarize_Luhn(input_2)
             tempSum3 = summarize_Luhn(input_3)
             fullSumm = tempSum2 + tempSum3
             finalSum = summarize_Luhn(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)     
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)     
     
     if method == "LexRank + mT5":
         if(input_1 and input_2 and input_3 ): #"3 cheios"
@@ -167,28 +167,28 @@ def summarize_HUB_Multidocument(input_1, input_2, input_3, method, max_length, n
             tempSum3 = summarize_LexRank(input_3)
             fullSumm = tempSum1 + tempSum2 + tempSum3
             finalSum = summarize_LexRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)   
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)   
             
         if(input_1 and input_2 and not input_3): #"1 e 2 cheios"
             tempSum1 = summarize_LexRank(input_1)
             tempSum2 = summarize_LexRank(input_2)
             fullSumm = tempSum1 + tempSum2
             finalSum = summarize_LexRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)     
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)     
         
         if(input_1 and not input_2 and input_3): #1 e 3 cheios"
             tempSum1 = summarize_LexRank(input_1)
             tempSum3 = summarize_LexRank(input_3)
             fullSumm = tempSum1 + tempSum3
             finalSum = summarize_LexRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)      
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)      
             
         if(not input_1 and input_2 and input_3): #"2 e 3 cheios"
             tempSum2 = summarize_LexRank(input_2)
             tempSum3 = summarize_LexRank(input_3)
             fullSumm = tempSum2 + tempSum3
             finalSum = summarize_LexRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)
     
     if method == "TextRank + mT5":
         if(input_1 and input_2 and input_3 ): #"3 cheios"
@@ -197,33 +197,33 @@ def summarize_HUB_Multidocument(input_1, input_2, input_3, method, max_length, n
             tempSum3 = summarize_TextRank(input_3)
             fullSumm = tempSum1 + tempSum2 + tempSum3
             finalSum = summarize_TextRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)   
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)   
             
         if(input_1 and input_2 and not input_3): #"1 e 2 cheios"
             tempSum1 = summarize_TextRank(input_1)
             tempSum2 = summarize_TextRank(input_2)
             fullSumm = tempSum1 + tempSum2
             finalSum = summarize_TextRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)     
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)     
         
         if(input_1 and not input_2 and input_3): #1 e 3 cheios"
             tempSum1 = summarize_TextRank(input_1)
             tempSum3 = summarize_TextRank(input_3)
             fullSumm = tempSum1 + tempSum3
             finalSum = summarize_TextRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)      
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)      
             
         if(not input_1 and input_2 and input_3): #"2 e 3 cheios"
             tempSum2 = summarize_TextRank(input_2)
             tempSum3 = summarize_TextRank(input_3)
             fullSumm = tempSum2 + tempSum3
             finalSum = summarize_TextRank(fullSumm)
-            return summarize_mT5(finalSum, max_length, num_beams)            
+            return summarize_mT5(finalSum, max_length, min_length, num_beams)            
     return "ERROR"
 
-def summarize_HUB_Monodocument(input, method, max_length, num_beams):
+def summarize_HUB_Monodocument(input, method, max_length, min_length, num_beams):
     if method == "Pure mT5":
-        return summarize_mT5(input, max_length, num_beams)
+        return summarize_mT5(input, max_length, min_length, num_beams)
     
     if method == "Luhn":
         return summarize_Luhn(input)
@@ -236,15 +236,15 @@ def summarize_HUB_Monodocument(input, method, max_length, num_beams):
     
     if method == "Luhn + mT5":
         tempSum = summarize_Luhn(input)
-        return summarize_mT5(tempSum, max_length, num_beams)
+        return summarize_mT5(tempSum, max_length, min_length, num_beams)
     
     if method == "LexRank + mT5":
         tempSum = summarize_LexRank(input)
-        return summarize_mT5(tempSum, max_length, num_beams)
+        return summarize_mT5(tempSum, max_length, min_length, num_beams)
     
     if method == "TextRank + mT5":
         tempSum = summarize_TextRank(input)
-        return summarize_mT5(tempSum, max_length, num_beams)
+        return summarize_mT5(tempSum, max_length, min_length, num_beams)
     return "ERROR"
 
 def summarize_Luhn(input):
@@ -280,7 +280,7 @@ def summarize_TextRank(input):
     summ2 = summ.replace('\n', ' ').replace('\r', '')
     return summ2;
 
-def summarize_mT5(input, max_length, num_beams):
+def summarize_mT5(input, max_length, min_length, num_beams):
   for i in range(0,14):
     input_ids = tokenizer(
         input,
@@ -293,7 +293,7 @@ def summarize_mT5(input, max_length, num_beams):
     output_ids = model.generate(
         input_ids=input_ids,
         max_length=max_length,
-        min_length=int(len(input.split())/3),
+        min_length=min_length,
         no_repeat_ngram_size=2,
         num_beams=num_beams
     )[0]
@@ -311,13 +311,13 @@ with app:
     
         with gr.TabItem("Sumarização Monodocumento"):
             MonoInputs=[gr.Textbox(label="Texto a ser Sumarizado"),gr.Radio(["Pure mT5","Luhn","LexRank","TextRank","Luhn + mT5","LexRank + mT5","TextRank + mT5"], label="Método"), 
-gr.Slider(50, 500, step=1, value=200, label="Tamanho máximo do Sumário"), gr.Slider(1, 10, step=1, value=4, label="Qualidade do sumário")]
+gr.Slider(50, 500, step=1, value=200, label="Tamanho máximo do Sumário"), gr.Slider(1, 125, step=1, value=50, label="Tamanho mínimo do Sumário"), gr.Slider(1, 10, step=1, value=4, label="Qualidade do sumário")]
             MonoOutputs=gr.Textbox()
             MonoButton = gr.Button("Sumarizar Texto")
             
         with gr.TabItem("Sumarização Multidocumento"):
             MultiInputs=[gr.Textbox(label="Texto 1"), gr.Textbox(label="Texto 2"),gr.Textbox(label="Texto 3"),gr.Radio(["Pure mT5","Luhn","LexRank","TextRank","Luhn + mT5","LexRank + mT5","TextRank + mT5"], label="Método"), 
-gr.Slider(50, 500, step=1, value=200, label="Tamanho máximo do Sumário"), gr.Slider(1, 10, step=1, value=4, label="Qualidade do sumário")]
+gr.Slider(50, 500, step=1, value=200, label="Tamanho máximo do Sumário"), gr.Slider(1, 125, step=1, value=50, label="Tamanho mínimo do Sumário"), gr.Slider(1, 10, step=1, value=4, label="Qualidade do sumário")]
             MultiOutputs=gr.Textbox()
             MultiButton = gr.Button("Sumarizar Textos")