Hugging Face's logo

Spaces:
ysharma
/
text_to_joke
Build error

App Files Community
text_to_joke
File size: 4,279 Bytes 
bcaa587
 
4439ec5
bcaa587
4901721
bcaa587
 
 
 
 
 
 
 
4a5c5dd
bcaa587
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5314799
bcaa587
 
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
 
import os
#os.system("pip install git+https://github.com/openai/whisper.git")
#os.system("pip install neon-tts-plugin-coqui==0.6.0") 
import gradio as gr
#import whisper 
import requests 
import tempfile
#from neon_tts_plugin_coqui import CoquiTTS
from datasets import load_dataset
import random

dataset = load_dataset("ysharma/short_jokes", split="train")
filtered_dataset = dataset.filter(
    lambda x: (True not in [nsfw in x["Joke"].lower() for nsfw in ["warning","porn", "blow", "fuck", "dead", "nsfw","69", "sex", "prostitute","prostitutes", "pedophiles", "pedophile"]]) 
    )


# Model 2: Sentence Transformer
API_URL = "https://api-inference.huggingface.co/models/sentence-transformers/msmarco-distilbert-base-tas-b"
HF_TOKEN = os.environ["HF_TOKEN"]
headers = {"Authorization": f"Bearer {HF_TOKEN}"}

def query(payload):
    response = requests.post(API_URL, headers=headers, json=payload)
    return response.json()


# Language common in both the multilingual models - English, Chinese, Spanish, and French etc
# Model 1: Whisper: Speech-to-text
#model = whisper.load_model("base")


#Model 2:  Text-to-Speech
#LANGUAGES = list(CoquiTTS.langs.keys())
#coquiTTS = CoquiTTS()
#Languages for Coqui are: ['en', 'es', 'fr', 'de', 'pl', 'uk', 'ro', 'hu', 'el', 'bg', 'nl', 'fi', 'sl', 'lv', 'ga']

  
# Driver function
def driver_fun(text) : 
  print("*********** Inside Driver ************")
  #if (text == 'dummy') and (audio is not None) :
  #  print(f"Audio is {audio}")
  #  translation, lang = whisper_stt(audio)  
  #else:
  #  translation = text 
    
  random_val = random.randrange(0,231657)
  if random_val < 226657:
    lower_limit = random_val
    upper_limit = random_val + 4000 
  else:
    lower_limit = random_val - 4000
    upper_limit = random_val 
  print(f"lower_limit : upper_limit = {lower_limit} : {upper_limit}")  
  dataset_subset = filtered_dataset['Joke'][lower_limit : upper_limit]
  data = query({"inputs": {"source_sentence": text ,"sentences": dataset_subset} } ) #"That is a happy person"
  if 'error' in data:
    print(f"Error is : {data}")
    return 'Error in model inference - Run Again Please', 'Error in model inference - Run Again Please', None
  print(f"type(data) : {type(data)}")
  #print(f"data : {data} ")
  max_match_score = max(data)
  indx_score = data.index(max_match_score)
  joke = dataset_subset[indx_score]
  print(f"Joke is : {joke}")
  
  #speech = tts(joke, 'en') 
  return joke 

demo = gr.Blocks()
with demo:
  gr.Markdown("<h1><center>Text-to-Joke</center></h1>")
  gr.Markdown(
        """<center>Enter a theme or a context for AI to find a joke for you on that.</center><br><center>If you see the message 'Error in model inference - Run Again Please', just press the button again every time!</center>
        """)
  with gr.Row():
    with gr.Column(): 
      #in_audio = gr.Audio(source="microphone",  type="filepath", label='Record your voice command here in English -')  #type='filepath'
      in_text = gr.Textbox(label= 'Enter a theme or context for a joke')
      b1 = gr.Button("Get a Joke")
      
    with gr.Column():
      #in_text = gr.Textbox(label='Or enter any text here..', value='dummy')
      #out_audio = gr.Audio(label='Audio response form CoquiTTS')  
      out_generated_joke = gr.Textbox(label= 'Joke returned! ')
      
    b1.click(driver_fun,inputs=[in_text], outputs=[out_generated_joke]) #out_translation_en, out_generated_text,out_generated_text_en, 
  with gr.Row():
    gr.Markdown(
        """Built using [Sentence Transformers](https://huggingface.co/models?library=sentence-transformers&sort=downloads) and [**Gradio Block API**](https://gradio.app/docs/#blocks).<br><br>Few Caveats:<br>1. Please note that sometimes the joke might be NSFW. Although, I have tried putting in filters to not have that experience, but the filters seem non-exhaustive.<br>2. Sometimes the joke might not match your theme, please bear with the limited capabilities of free open-source ML prototypes.<br>3. Much like real life, sometimes the joke might just not land, haha!<br>4. Repeating this: If you see the message 'Error in model inference - Run Again Please', just press the button again every time!
        """)
demo.queue(concurrency_count=3)    
demo.launch(enable_queue=True, debug=True)