added rhoaplisms and E-prime setting

app.py

@@ -1,6 +1,6 @@
 import re
 from unittest import result
-
+import string
 import streamlit as st
 import torch
 from torch.nn import functional as F
@@ -25,15 +25,23 @@ form.header("Main Settings")
 
 model_name = form.text_area("Enter the name of the pre-trained model from transformers that we are using for Text Generation", value = "gpt2")
 form.caption("This will download a new model, so it may take awhile or even break if the model is too large")
-mode = form.selectbox("What kind of constrained generation are we doing?", ["lipogram", "reverse_lipogram", "length_constrained", "greater_than_length"])
+mode = form.selectbox("What kind of constrained generation are we doing?", ["lipogram", "reverse_lipogram", "e-prime", "rhopalism", "length_constrained", "greater_than_length"])
 form.caption("Lipograms mean that a letter (or substring) is not allowed in the generated string, reverse lipograms force a letter to be in the generated string")
 
 if mode == "lipogram":
     naughty_strings_list = st.text_area("Enter the list of strings that you don't want in each word seperated by a space", value = "E e")
     naughty_strings = naughty_strings_list.split(" ")
+elif mode == "e-prime":
+    e_prime_string = """be being been am is isn't are aren't was wasn't were weren't i'm you're we're they're he's she's it's there's here's where's how's what's who's that's aint isnt arent wasnt werent im youre were theyre hes shes its theres heres wheres hows whats whos thats aint Be Being Been Am Is Isn't Are Aren't Was Wasn't Were Weren't I'm You're We're They're He's She's It's There's Here's Where's How's What's Who's That's Aint Isnt Arent Wasnt Werent Im Youre Were Theyre Hes Shes Its Theres Heres Wheres Hows Whats Whos Thats Aint BE BEING BEEN AM IS ISN'T ARE AREN'T WAS WASN'T WERE WEREN'T I'M YOU'RE WE'RE THEY'RE HE'S SHE'S IT'S THERE'S HERE'S WHERE'S HOW'S WHAT'S WHO'S THAT'S AINT ISNT ARENT WASNT WERENT IM YOURE WERE THEYRE HES SHES ITS THERES HERES WHERES HOWS WHATS WHOS THATS AINT"""
+    st.caption("The default word list is the list needed to enforce the language model to generate english without usage of the verb to be")
+    naughty_strings_list = st.text_area("Enter the list of strings that you don't want to be generated (exact match)", value = e_prime_string)
+    naughty_strings = naughty_strings_list.split(" ")    
 elif mode == "reverse_lipogram":
     nice_strings_list = st.text_area("Enter the list of strings that you DO want in each word seperated by a space", value = "t T")
     nice_strings = nice_strings_list.split(" ")
+elif mode == "rhopalism":
+    length_constraint = form.number_input("Enter the length that the Rhopalism shoud start with", value = 1)
+    st.caption("Rhopalisms are usually reliable but sometimes you need to try generating two or three times for a perfect one")
 else:
     length_constraint = form.number_input("Enter the length should each word be restricted to (or greater/less than)", value = 5) + 1
 
@@ -59,6 +67,9 @@ with st.spinner("Please wait while the model loads:"):
 def isPalindrome(s):
     return s == s[::-1]
 
+rhopalism_len = length_constraint
+alphabet_string = string.ascii_lowercase
+
 
 def get_next_word_without_e(input_sequence):
     input_ids = tokenizer.encode(sequence, return_tensors="pt")
@@ -77,7 +88,7 @@ def get_next_word_without_e(input_sequence):
             resulting_string = tokenizer.decode(candidate) #skip_special_tokens=True, clean_up_tokenization_spaces=True)
             ###Constrained text generation starts HERE
             ##Lipogram - No naughty strings used
-            if mode == "lipogram":
+            if mode == "lipogram" or mode == "e-prime":
                 if all(nauty_string not in resulting_string for nauty_string in naughty_strings): ## This returns at the first naughty strings
                     return resulting_string
             ##Reverse-Lipogram - Must use things in nice_strings
@@ -93,6 +104,10 @@ def get_next_word_without_e(input_sequence):
                 ##Only sort of works 
                 if len(resulting_string) >= length_constraint:
                     return resulting_string          
+            elif mode == "rhopalism":
+                ##Mostly works
+                if len(resulting_string) == rhopalism_len:
+                    return resulting_string
     return " "
 
 
@@ -101,6 +116,7 @@ i = length
 while i > 0:
     new_word = get_next_word_without_e(input_sequence= sequence)
     sequence = sequence + new_word
+    rhopalism_len += 1
     i = i-1
 
 st.write("GENERATED SEQUENCE: ")