Spaces:
Build error
Build error
File size: 11,772 Bytes
978ee76 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 |
import gradio as gr
from transformers import (
pipeline,
AutoModelForSeq2SeqLM,
AutoTokenizer
)
M1 = "anshoomehra/question-generation-auto-t5-v1-base-s-q"
M2 = "anshoomehra/question-generation-auto-t5-v1-base-s-q-c"
M4 = "anshoomehra/question-generation-auto-hints-t5-v1-base-s-q"
M5 = "anshoomehra/question-generation-auto-hints-t5-v1-base-s-q-c"
device = ['cuda' if torch.cuda.is_available() else 'cpu'][0]
_m1 = AutoModelForSeq2SeqLM.from_pretrained(M1).to(device)
_tk1 = AutoTokenizer.from_pretrained(M1, cache_dir="./cache")
_m2 = AutoModelForSeq2SeqLM.from_pretrained(M2).to(device)
_tk2 = AutoTokenizer.from_pretrained(M2, cache_dir="./cache")
_m4 = AutoModelForSeq2SeqLM.from_pretrained(M4).to(device)
_tk4 = AutoTokenizer.from_pretrained(M4, cache_dir="./cache")
_m5 = AutoModelForSeq2SeqLM.from_pretrained(M5).to(device)
_tk5 = AutoTokenizer.from_pretrained(M5, cache_dir="./cache")
def _formatQs(questions):
_finalQs = ""
if questions is not None:
_qList = questions[0].strip().split("?")
qIdx = 1
if len(_qList) > 1:
for idx, _q in enumerate(_qList):
_q = _q.strip()
if _q is not None and len(_q) !=0:
_finalQs += str(qIdx) + ". " + _q + "? \n"
qIdx+=1
else:
if len(_qList[0])>1:
_finalQs = "1. " + str(_qList[0]) + "?"
else:
_finalQs = None
return _finalQs
def _generate(mode, context, hint=None, minLength=50, maxLength=500, lengthPenalty=2.0, earlyStopping=True, numReturnSequences=1, numBeams=2, noRepeatNGramSize=0, doSample=False, topK=0, topP=0, temperature=0):
predictionM1 = None
predictionM2 = None
predictionM4 = None
predictionM5 = None
if mode == 'Auto':
_inputText = "question_context: " + context
_encoding = _tk1.encode(_inputText, return_tensors='pt', truncation=True, padding='max_length').to(device) # max_length=1024
_outputEncoded = _m1.generate(_encoding,
min_length=minLength,
max_length=maxLength,
length_penalty=lengthPenalty,
early_stopping=earlyStopping,
num_return_sequences=numReturnSequences,
num_beams=numBeams,
no_repeat_ngram_size=noRepeatNGramSize,
do_sample=doSample,
top_k=topK,
top_p=topP,
temperature=temperature
)
predictionM1 = [_tk1.decode(id, clean_up_tokenization_spaces=False, skip_special_tokens=True) for id in _outputEncoded]
_encoding = _tk2.encode(_inputText, return_tensors='pt', truncation=True, padding='max_length').to(device) # max_length=1024 .to(device)
_outputEncoded = _m2.generate(_encoding,
min_length=minLength,
max_length=maxLength,
length_penalty=lengthPenalty,
early_stopping=earlyStopping,
num_return_sequences=numReturnSequences,
num_beams=numBeams,
no_repeat_ngram_size=noRepeatNGramSize,
do_sample=doSample,
top_k=topK,
top_p=topP,
temperature=temperature
)
predictionM2 = [_tk2.decode(id, clean_up_tokenization_spaces=False, skip_special_tokens=True) for id in _outputEncoded]
_encoding = _tk4.encode(_inputText, return_tensors='pt', truncation=True, padding='max_length').to(device) # max_length=1024 .to(device)
_outputEncoded = _m4.generate(_encoding,
min_length=minLength,
max_length=maxLength,
length_penalty=lengthPenalty,
early_stopping=earlyStopping,
num_return_sequences=numReturnSequences,
num_beams=numBeams,
no_repeat_ngram_size=noRepeatNGramSize,
do_sample=doSample,
top_k=topK,
top_p=topP,
temperature=temperature
)
predictionM4 = [_tk4.decode(id, clean_up_tokenization_spaces=False, skip_special_tokens=True) for id in _outputEncoded]
_encoding = _tk5.encode(_inputText, return_tensors='pt', truncation=True, padding='max_length').to(device) # max_length=1024 .to(device)
_outputEncoded = _m5.generate(_encoding,
min_length=minLength,
max_length=maxLength,
length_penalty=lengthPenalty,
early_stopping=earlyStopping,
num_return_sequences=numReturnSequences,
num_beams=numBeams,
no_repeat_ngram_size=noRepeatNGramSize,
do_sample=doSample,
top_k=topK,
top_p=topP,
temperature=temperature
)
predictionM5 = [_tk5.decode(id, clean_up_tokenization_spaces=False, skip_special_tokens=True) for id in _outputEncoded]
elif mode == 'Hints':
_inputText = "question_hint: " + hint + "</s>question_context: " + context
_encoding = _tk4.encode(_inputText, return_tensors='pt', truncation=True, padding='max_length').to(device) # max_length=1024 .to(device)
_outputEncoded = _m4.generate(_encoding,
min_length=minLength,
max_length=maxLength,
length_penalty=lengthPenalty,
early_stopping=earlyStopping,
num_return_sequences=numReturnSequences,
num_beams=numBeams,
no_repeat_ngram_size=noRepeatNGramSize,
do_sample=doSample,
top_k=topK,
top_p=topP,
temperature=temperature
)
predictionM4 = [_tk4.decode(id, clean_up_tokenization_spaces=False, skip_special_tokens=True) for id in _outputEncoded]
_encoding = _tk5.encode(_inputText, return_tensors='pt', truncation=True, padding='max_length').to(device) # max_length=1024 .to(device)
_outputEncoded = _m5.generate(_encoding,
min_length=minLength,
max_length=maxLength,
length_penalty=lengthPenalty,
early_stopping=earlyStopping,
num_return_sequences=numReturnSequences,
num_beams=numBeams,
no_repeat_ngram_size=noRepeatNGramSize,
do_sample=doSample,
top_k=topK,
top_p=topP,
temperature=temperature
)
predictionM5 = [_tk5.decode(id, clean_up_tokenization_spaces=False, skip_special_tokens=True) for id in _outputEncoded]
predictionM1 = _formatQs(predictionM1)
predictionM2 = _formatQs(predictionM2)
predictionM4 = _formatQs(predictionM4)
predictionM5 = _formatQs(predictionM5)
return predictionM1, predictionM2, predictionM4, predictionM5
with gr.Blocks() as demo:
gr.Markdown(value="# Question Generation Demo")
with gr.Accordion(variant='compact', label='Hyperparams', open=False):
with gr.Row():
mode = gr.Radio(["Auto", "Hints"], value="Auto", label="Mode")
with gr.Row():
minLength = gr.Slider(10, 512, 50, step=1, label="Min Length")
maxLength = gr.Slider(20, 512, 164, step=1, label="Max Length")
lengthPenalty = gr.Slider(-5, 5, 1, label="Length Penalty")
earlyStopping = gr.Checkbox(True, label="Early Stopping [EOS]")
numReturnSequences = gr.Slider(1, 3, 1, step=1, label="Num return Sequences")
with gr.Row():
numBeams = gr.Slider(1, 10, 4, step=1, label="Beams")
noRepeatNGramSize = gr.Slider(0, 5, 3, step=1, label="No Repeat N-Gram Size")
with gr.Row():
doSample = gr.Checkbox(label="Do Random Sample")
topK = gr.Slider(0, 50, 0, step=1, label="Top K")
topP = gr.Slider(0, 1, 0, label="Top P/Nucleus Sampling")
temperature = gr.Slider(0.01, 1, 1, label="Temperature")
with gr.Accordion(variant='compact', label='Input Values'):
with gr.Row(variant='compact'):
contextDefault = "Google LLC is an American multinational technology company focusing on search engine technology, online advertising, cloud computing, computer software, quantum computing, e-commerce, artificial intelligence, and consumer electronics. It has been referred to as 'the most powerful company in the world' and one of the world's most valuable brands due to its market dominance, data collection, and technological advantages in the area of artificial intelligence. Its parent company Alphabet is considered one of the Big Five American information technology companies, alongside Amazon, Apple, Meta, and Microsoft."
hintDefault = ""
context = gr.Textbox(contextDefault, label="Context", placeholder="Dummy Context", lines=5)
hint = gr.Textbox(hintDefault, label="Hint", placeholder="Enter hint here. Ensure the mode is set to 'Hints' prior using hints.", lines=2)
with gr.Accordion(variant='compact', label='Multi-Task Model(s) Sensitive To Hints'):
with gr.Row(variant='compact'):
_predictionM5 = gr.Textbox(label="Predicted Questions - question-generation-auto-hints-t5-v1-base-s-q-c [Hints Sensitive]")
_predictionM4 = gr.Textbox(label="Predicted Questions - question-generation-auto-hints-t5-v1-base-s-q [Hints Sensitive]")
with gr.Accordion(variant='compact', label='Uni-Task Model(s) Non-Sensitive To Hints'):
with gr.Row(variant='compact'):
_predictionM2 = gr.Textbox(label="Predicted Questions - question-generation-auto-t5-v1-base-s-q-c [No Hints]")
_predictionM1 = gr.Textbox(label="Predicted Questions - question-generation-auto-t5-v1-base-s-q [No Hints]")
with gr.Row():
gen_btn = gr.Button("Generate Questions")
gen_btn.click(fn=_generate,
inputs=[mode, context, hint, minLength, maxLength, lengthPenalty, earlyStopping, numReturnSequences, numBeams, noRepeatNGramSize, doSample, topK, topP, temperature],
outputs=[_predictionM1, _predictionM2, _predictionM4, _predictionM5]
) |