JohnTony commited on
Commit
4cae33d
1 Parent(s): 13dbf36

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +270 -0
app.py ADDED
@@ -0,0 +1,270 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import streamlit as st
2
+ import numpy as np
3
+ import pandas as pd
4
+ import os
5
+ import torch
6
+ import torch.nn as nn
7
+ from transformers.activations import get_activation
8
+ from transformers import AutoTokenizer, AutoModelForCausalLM
9
+
10
+
11
+ st.title('GPT2: To see all prompt outlines: https://huggingface.co/BigSalmon/InformalToFormalLincoln64Paraphrase')
12
+
13
+ device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
14
+
15
+ @st.cache(allow_output_mutation=True)
16
+ def get_model():
17
+
18
+ tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln91Paraphrase")
19
+ model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln91Paraphrase")
20
+
21
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln90Paraphrase")
22
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln90Paraphrase")
23
+
24
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln88Paraphrase")
25
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln88Paraphrase")
26
+
27
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln86Paraphrase")
28
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln86Paraphrase")
29
+
30
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln82Paraphrase")
31
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln82Paraphrase")
32
+
33
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln79Paraphrase")
34
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln79Paraphrase")
35
+
36
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln74Paraphrase")
37
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln74Paraphrase")
38
+
39
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln72Paraphrase")
40
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln72Paraphrase")
41
+
42
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln64Paraphrase")
43
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln64Paraphrase")
44
+
45
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln60Paraphrase")
46
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln60Paraphrase")
47
+
48
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/GPTNeo1.3BInformalToFormal")
49
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/GPTNeo1.3BInformalToFormal")
50
+
51
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln55")
52
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln55")
53
+
54
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln51")
55
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln51")
56
+
57
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln45")
58
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln49")
59
+
60
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln43")
61
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln43")
62
+
63
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln41")
64
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln41")
65
+
66
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln38")
67
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln38")
68
+
69
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln37")
70
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln37")
71
+
72
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln36")
73
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln36")
74
+
75
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/MediumInformalToFormalLincoln")
76
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/MediumInformalToFormalLincoln")
77
+
78
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln35")
79
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln35")
80
+
81
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln31")
82
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln31")
83
+
84
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/InformalToFormalLincoln21")
85
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/InformalToFormalLincoln21")
86
+
87
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/PointsOneSent")
88
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/PointsOneSent")
89
+
90
+ #tokenizer = AutoTokenizer.from_pretrained("BigSalmon/PointsToSentence")
91
+ #model = AutoModelForCausalLM.from_pretrained("BigSalmon/PointsToSentence")
92
+
93
+ return model, tokenizer
94
+
95
+ model, tokenizer = get_model()
96
+
97
+ g = """informal english: garage band has made people who know nothing about music good at creating music.
98
+ Translated into the Style of Abraham Lincoln: garage band ( offers the uninitiated in music the ability to produce professional-quality compositions / catapults those for whom music is an uncharted art the ability the realize masterpieces / stimulates music novice's competency to yield sublime arrangements / begets individuals of rudimentary musical talent the proficiency to fashion elaborate suites ).
99
+ informal english: chrome extensions can make doing regular tasks much easier to get done.
100
+ Translated into the Style of Abraham Lincoln: chrome extensions ( yield the boon of time-saving convenience / ( expedite the ability to / unlock the means to more readily ) accomplish everyday tasks / turbocharges the velocity with which one can conduct their obligations ).
101
+ informal english: broadband is finally expanding to rural areas, a great development that will thrust them into modern life.
102
+ Translated into the Style of Abraham Lincoln: broadband is ( ( finally / at last / after years of delay ) arriving in remote locations / springing to life in far-flung outposts / inching into even the most backwater corners of the nation ) that will leap-frog them into the twenty-first century.
103
+ informal english: google translate has made talking to people who do not share your language easier.
104
+ Translated into the Style of Abraham Lincoln: google translate ( imparts communicability to individuals whose native tongue differs / mitigates the trials of communication across linguistic barriers / hastens the bridging of semantic boundaries / mollifies the complexity of multilingual communication / avails itself to the internationalization of discussion / flexes its muscles to abet intercultural conversation / calms the tides of linguistic divergence ).
105
+ informal english: corn fields are all across illinois, visible once you leave chicago.
106
+ Translated into the Style of Abraham Lincoln: corn fields ( permeate illinois / span the state of illinois / ( occupy / persist in ) all corners of illinois / line the horizon of illinois / envelop the landscape of illinois ), manifesting themselves visibly as one ventures beyond chicago.
107
+ informal english: """
108
+
109
+ number_of_outputs = st.sidebar.slider("Number of Outputs", 5, 100)
110
+ log_nums = st.sidebar.slider("How Many Log Outputs?", 50, 600)
111
+
112
+ def BestProbs(prompt):
113
+ prompt = prompt.strip()
114
+ text = tokenizer.encode(prompt)
115
+ myinput, past_key_values = torch.tensor([text]), None
116
+ myinput = myinput
117
+ logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)
118
+ logits = logits[0,-1]
119
+ probabilities = torch.nn.functional.softmax(logits)
120
+ best_logits, best_indices = logits.topk(10)
121
+ best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]
122
+ for i in best_words[0:10]:
123
+ print("_______")
124
+ st.write(f"${i} $\n")
125
+ f = (f"${i} $\n")
126
+ m = (prompt + f"{i}")
127
+ BestProbs2(m)
128
+ return f
129
+
130
+ def BestProbs2(prompt):
131
+ prompt = prompt.strip()
132
+ text = tokenizer.encode(prompt)
133
+ myinput, past_key_values = torch.tensor([text]), None
134
+ myinput = myinput
135
+ logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)
136
+ logits = logits[0,-1]
137
+ probabilities = torch.nn.functional.softmax(logits)
138
+ best_logits, best_indices = logits.topk(20)
139
+ best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]
140
+ for i in best_words[0:20]:
141
+ print(i)
142
+ st.write(i)
143
+
144
+ def LogProbs(prompt):
145
+ col1 = []
146
+ col2 = []
147
+ prompt = prompt.strip()
148
+ text = tokenizer.encode(prompt)
149
+ myinput, past_key_values = torch.tensor([text]), None
150
+ myinput = myinput
151
+ logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)
152
+ logits = logits[0,-1]
153
+ probabilities = torch.nn.functional.softmax(logits)
154
+ best_logits, best_indices = logits.topk(10)
155
+ best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]
156
+ for i in best_words[0:10]:
157
+ print("_______")
158
+ f = i
159
+ col1.append(f)
160
+ m = (prompt + f"{i}")
161
+ #print("^^" + f + " ^^")
162
+ prompt = m.strip()
163
+ text = tokenizer.encode(prompt)
164
+ myinput, past_key_values = torch.tensor([text]), None
165
+ myinput = myinput
166
+ logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)
167
+ logits = logits[0,-1]
168
+ probabilities = torch.nn.functional.softmax(logits)
169
+ best_logits, best_indices = logits.topk(20)
170
+ best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]
171
+ for i in best_words[0:20]:
172
+ #print(i)
173
+ col2.append(i)
174
+ #print(col1)
175
+ #print(col2)
176
+ d = {col1[0]: [col2[0], col2[1], col2[2], col2[3], col2[4], col2[5], col2[6], col2[7], col2[8], col2[9], col2[10], col2[11], col2[12], col2[13], col2[14], col2[15], col2[16], col2[17], col2[18], col2[19]],
177
+ col1[1]: [col2[20], col2[21], col2[22], col2[23], col2[24], col2[25], col2[26], col2[27], col2[28], col2[29], col2[30], col2[31], col2[32], col2[33], col2[34], col2[35], col2[36], col2[37], col2[38], col2[39]],
178
+ col1[2]: [col2[40], col2[41], col2[42], col2[43], col2[44], col2[45], col2[46], col2[47], col2[48], col2[49], col2[50], col2[51], col2[52], col2[53], col2[54], col2[55], col2[56], col2[57], col2[58], col2[59]],
179
+ col1[3]: [col2[60], col2[61], col2[62], col2[63], col2[64], col2[65], col2[66], col2[67], col2[68], col2[69], col2[70], col2[71], col2[72], col2[73], col2[74], col2[75], col2[76], col2[77], col2[78], col2[79]],
180
+ col1[4]: [col2[80], col2[81], col2[82], col2[83], col2[84], col2[85], col2[86], col2[87], col2[88], col2[89], col2[90], col2[91], col2[92], col2[93], col2[94], col2[95], col2[96], col2[97], col2[98], col2[99]],
181
+ col1[5]: [col2[100], col2[101], col2[102], col2[103], col2[104], col2[105], col2[106], col2[107], col2[108], col2[109], col2[110], col2[111], col2[112], col2[113], col2[114], col2[115], col2[116], col2[117], col2[118], col2[119]],
182
+ col1[6]: [col2[120], col2[121], col2[122], col2[123], col2[124], col2[125], col2[126], col2[127], col2[128], col2[129], col2[130], col2[131], col2[132], col2[133], col2[134], col2[135], col2[136], col2[137], col2[138], col2[139]],
183
+ col1[7]: [col2[140], col2[141], col2[142], col2[143], col2[144], col2[145], col2[146], col2[147], col2[148], col2[149], col2[150], col2[151], col2[152], col2[153], col2[154], col2[155], col2[156], col2[157], col2[158], col2[159]],
184
+ col1[8]: [col2[160], col2[161], col2[162], col2[163], col2[164], col2[165], col2[166], col2[167], col2[168], col2[169], col2[170], col2[171], col2[172], col2[173], col2[174], col2[175], col2[176], col2[177], col2[178], col2[179]],
185
+ col1[9]: [col2[180], col2[181], col2[182], col2[183], col2[184], col2[185], col2[186], col2[187], col2[188], col2[189], col2[190], col2[191], col2[192], col2[193], col2[194], col2[195], col2[196], col2[197], col2[198], col2[199]]}
186
+ df = pd.DataFrame(data=d)
187
+ print(df)
188
+ st.write(df)
189
+ return df
190
+
191
+ def BestProbs5(prompt):
192
+ prompt = prompt.strip()
193
+ text = tokenizer.encode(prompt)
194
+ myinput, past_key_values = torch.tensor([text]), None
195
+ myinput = myinput
196
+ logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)
197
+ logits = logits[0,-1]
198
+ probabilities = torch.nn.functional.softmax(logits)
199
+ best_logits, best_indices = logits.topk(number_of_outputs)
200
+ best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]
201
+ for i in best_words[0:number_of_outputs]:
202
+ #print(i)
203
+ print("\n")
204
+ g = (prompt + i)
205
+ st.write(g)
206
+ l = run_generate(g, "hey")
207
+ st.write(l)
208
+
209
+ def run_generate(text, bad_words):
210
+ yo = []
211
+ input_ids = tokenizer.encode(text, return_tensors='pt')
212
+ res = len(tokenizer.encode(text))
213
+ bad_words = bad_words.split()
214
+ bad_word_ids = [[7829], [40940]]
215
+ for bad_word in bad_words:
216
+ bad_word = " " + bad_word
217
+ ids = tokenizer(bad_word).input_ids
218
+ bad_word_ids.append(ids)
219
+ sample_outputs = model.generate(
220
+ input_ids,
221
+ do_sample=True,
222
+ max_length= res + 5,
223
+ min_length = res + 5,
224
+ top_k=50,
225
+ temperature=1.0,
226
+ num_return_sequences=3,
227
+ bad_words_ids=bad_word_ids
228
+ )
229
+ for i in range(3):
230
+ e = tokenizer.decode(sample_outputs[i])
231
+ e = e.replace(text, "")
232
+ yo.append(e)
233
+ print(yo)
234
+ return yo
235
+
236
+ with st.form(key='my_form'):
237
+ prompt = st.text_area(label='Enter sentence', value=g, height=500)
238
+ submit_button = st.form_submit_button(label='Submit')
239
+ submit_button2 = st.form_submit_button(label='Fast Forward')
240
+ submit_button3 = st.form_submit_button(label='Fast Forward 2.0')
241
+ submit_button4 = st.form_submit_button(label='Get Top')
242
+
243
+ if submit_button:
244
+ with torch.no_grad():
245
+ text = tokenizer.encode(prompt)
246
+ myinput, past_key_values = torch.tensor([text]), None
247
+ myinput = myinput
248
+ myinput= myinput.to(device)
249
+ logits, past_key_values = model(myinput, past_key_values = past_key_values, return_dict=False)
250
+ logits = logits[0,-1]
251
+ probabilities = torch.nn.functional.softmax(logits)
252
+ best_logits, best_indices = logits.topk(log_nums)
253
+ best_words = [tokenizer.decode([idx.item()]) for idx in best_indices]
254
+ text.append(best_indices[0].item())
255
+ best_probabilities = probabilities[best_indices].tolist()
256
+ words = []
257
+ st.write(best_words)
258
+ if submit_button2:
259
+ print("----")
260
+ st.write("___")
261
+ m = LogProbs(prompt)
262
+ st.write("___")
263
+ st.write(m)
264
+ st.write("___")
265
+ if submit_button3:
266
+ print("----")
267
+ st.write("___")
268
+ st.write(BestProbs)
269
+ if submit_button4:
270
+ BestProbs5(prompt)