furqankassa commited on
Commit
8492787
1 Parent(s): 946a554

Upload 2 files

Browse files
Files changed (2) hide show
  1. app (7).py +209 -0
  2. requirements (7).txt +6 -0
app (7).py ADDED
@@ -0,0 +1,209 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import gradio as gr
2
+ import os
3
+ import json
4
+ import requests
5
+
6
+ #Streaming endpoint
7
+ API_URL = "https://api.openai.com/v1/chat/completions" #os.getenv("API_URL") + "/generate_stream"
8
+ OPENAI_API_KEY= os.environ["HF_TOKEN"] # Add a token to this space . Then copy it to the repository secret in this spaces settings panel. os.environ reads from there.
9
+ # Keys for Open AI ChatGPT API usage are created from here: https://platform.openai.com/account/api-keys
10
+
11
+ def predict(inputs, top_p, temperature, chat_counter, chatbot=[], history=[]): #repetition_penalty, top_k
12
+
13
+ # 1. Set up a payload
14
+ payload = {
15
+ "model": "gpt-3.5-turbo",
16
+ "messages": [{"role": "user", "content": f"{inputs}"}],
17
+ "temperature" : 1.0,
18
+ "top_p":1.0,
19
+ "n" : 1,
20
+ "stream": True,
21
+ "presence_penalty":0,
22
+ "frequency_penalty":0,
23
+ }
24
+
25
+ # 2. Define your headers and add a key from https://platform.openai.com/account/api-keys
26
+ headers = {
27
+ "Content-Type": "application/json",
28
+ "Authorization": f"Bearer {OPENAI_API_KEY}"
29
+ }
30
+
31
+ # 3. Create a chat counter loop that feeds [Predict next best anything based on last input and attention with memory defined by introspective attention over time]
32
+ print(f"chat_counter - {chat_counter}")
33
+ if chat_counter != 0 :
34
+ messages=[]
35
+ for data in chatbot:
36
+ temp1 = {}
37
+ temp1["role"] = "user"
38
+ temp1["content"] = data[0]
39
+ temp2 = {}
40
+ temp2["role"] = "assistant"
41
+ temp2["content"] = data[1]
42
+ messages.append(temp1)
43
+ messages.append(temp2)
44
+ temp3 = {}
45
+ temp3["role"] = "user"
46
+ temp3["content"] = inputs
47
+ messages.append(temp3)
48
+ payload = {
49
+ "model": "gpt-3.5-turbo",
50
+ "messages": messages, #[{"role": "user", "content": f"{inputs}"}],
51
+ "temperature" : temperature, #1.0,
52
+ "top_p": top_p, #1.0,
53
+ "n" : 1,
54
+ "stream": True,
55
+ "presence_penalty":0,
56
+ "frequency_penalty":0,
57
+ }
58
+ chat_counter+=1
59
+
60
+ # 4. POST it to OPENAI API
61
+ history.append(inputs)
62
+ print(f"payload is - {payload}")
63
+ response = requests.post(API_URL, headers=headers, json=payload, stream=True)
64
+ token_counter = 0
65
+ partial_words = ""
66
+
67
+ # 5. Iterate through response lines and structure readable response
68
+ counter=0
69
+ for chunk in response.iter_lines():
70
+ if counter == 0:
71
+ counter+=1
72
+ continue
73
+ if chunk.decode() :
74
+ chunk = chunk.decode()
75
+ if len(chunk) > 12 and "content" in json.loads(chunk[6:])['choices'][0]['delta']:
76
+ partial_words = partial_words + json.loads(chunk[6:])['choices'][0]["delta"]["content"]
77
+ if token_counter == 0:
78
+ history.append(" " + partial_words)
79
+ else:
80
+ history[-1] = partial_words
81
+ chat = [(history[i], history[i + 1]) for i in range(0, len(history) - 1, 2) ] # convert to tuples of list
82
+ token_counter+=1
83
+ yield chat, history, chat_counter
84
+
85
+
86
+ def reset_textbox():
87
+ return gr.update(value='')
88
+
89
+
90
+
91
+
92
+ # Episodic and Semantic IO
93
+ def list_files(file_path):
94
+ import os
95
+ icon_csv = "📄 "
96
+ icon_txt = "📑 "
97
+ current_directory = os.getcwd()
98
+ file_list = []
99
+ for filename in os.listdir(current_directory):
100
+ if filename.endswith(".csv"):
101
+ file_list.append(icon_csv + filename)
102
+ elif filename.endswith(".txt"):
103
+ file_list.append(icon_txt + filename)
104
+ if file_list:
105
+ return "\n".join(file_list)
106
+ else:
107
+ return "No .csv or .txt files found in the current directory."
108
+
109
+ # Function to read a file
110
+ def read_file(file_path):
111
+ try:
112
+ with open(file_path, "r") as file:
113
+ contents = file.read()
114
+ return f"{contents}"
115
+ #return f"Contents of {file_path}:\n{contents}"
116
+ except FileNotFoundError:
117
+ return "File not found."
118
+
119
+ # Function to delete a file
120
+ def delete_file(file_path):
121
+ try:
122
+ import os
123
+ os.remove(file_path)
124
+ return f"{file_path} has been deleted."
125
+ except FileNotFoundError:
126
+ return "File not found."
127
+
128
+ # Function to write to a file
129
+ def write_file(file_path, content):
130
+ try:
131
+ with open(file_path, "w") as file:
132
+ file.write(content)
133
+ return f"Successfully written to {file_path}."
134
+ except:
135
+ return "Error occurred while writing to file."
136
+
137
+ # Function to append to a file
138
+ def append_file(file_path, content):
139
+ try:
140
+ with open(file_path, "a") as file:
141
+ file.write(content)
142
+ return f"Successfully appended to {file_path}."
143
+ except:
144
+ return "Error occurred while appending to file."
145
+
146
+
147
+ title = """<h1 align="center">Memory Chat Story Generator ChatGPT</h1>"""
148
+ description = """
149
+ ## ChatGPT Datasets 📚
150
+ - WebText
151
+ - Common Crawl
152
+ - BooksCorpus
153
+ - English Wikipedia
154
+ - Toronto Books Corpus
155
+ - OpenWebText
156
+ ## ChatGPT Datasets - Details 📚
157
+ - **WebText:** A dataset of web pages crawled from domains on the Alexa top 5,000 list. This dataset was used to pretrain GPT-2.
158
+ - [WebText: A Large-Scale Unsupervised Text Corpus by Radford et al.](https://paperswithcode.com/dataset/webtext)
159
+ - **Common Crawl:** A dataset of web pages from a variety of domains, which is updated regularly. This dataset was used to pretrain GPT-3.
160
+ - [Language Models are Few-Shot Learners](https://paperswithcode.com/dataset/common-crawl) by Brown et al.
161
+ - **BooksCorpus:** A dataset of over 11,000 books from a variety of genres.
162
+ - [Scalable Methods for 8 Billion Token Language Modeling](https://paperswithcode.com/dataset/bookcorpus) by Zhu et al.
163
+ - **English Wikipedia:** A dump of the English-language Wikipedia as of 2018, with articles from 2001-2017.
164
+ - [Improving Language Understanding by Generative Pre-Training](https://huggingface.co/spaces/awacke1/WikipediaUltimateAISearch?logs=build) Space for Wikipedia Search
165
+ - **Toronto Books Corpus:** A dataset of over 7,000 books from a variety of genres, collected by the University of Toronto.
166
+ - [Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond](https://paperswithcode.com/dataset/bookcorpus) by Schwenk and Douze.
167
+ - **OpenWebText:** A dataset of web pages that were filtered to remove content that was likely to be low-quality or spammy. This dataset was used to pretrain GPT-3.
168
+ - [Language Models are Few-Shot Learners](https://paperswithcode.com/dataset/openwebtext) by Brown et al.
169
+ """
170
+
171
+ # 6. Use Gradio to pull it all together
172
+ with gr.Blocks(css = """#col_container {width: 1400px; margin-left: auto; margin-right: auto;} #chatbot {height: 600px; overflow: auto;}""") as demo:
173
+ gr.HTML(title)
174
+ with gr.Column(elem_id = "col_container"):
175
+ inputs = gr.Textbox(placeholder= "Hi there!", label= "Type an input and press Enter")
176
+ chatbot = gr.Chatbot(elem_id='chatbot')
177
+ state = gr.State([])
178
+ b1 = gr.Button()
179
+ with gr.Accordion("Parameters", open=False):
180
+ top_p = gr.Slider( minimum=-0, maximum=1.0, value=1.0, step=0.05, interactive=True, label="Top-p (nucleus sampling)",)
181
+ temperature = gr.Slider( minimum=-0, maximum=5.0, value=1.0, step=0.1, interactive=True, label="Temperature",)
182
+ chat_counter = gr.Number(value=0, visible=True, precision=0)
183
+
184
+
185
+ # Episodic/Semantic IO
186
+ fileName = gr.Textbox(label="Filename")
187
+ fileContent = gr.TextArea(label="File Content")
188
+ completedMessage = gr.Textbox(label="Completed")
189
+ label = gr.Label()
190
+ with gr.Row():
191
+ listFiles = gr.Button("📄 List File(s)")
192
+ readFile = gr.Button("📖 Read File")
193
+ saveFile = gr.Button("💾 Save File")
194
+ deleteFile = gr.Button("🗑️ Delete File")
195
+ appendFile = gr.Button("➕ Append File")
196
+ listFiles.click(list_files, inputs=fileName, outputs=fileContent)
197
+ readFile.click(read_file, inputs=fileName, outputs=fileContent)
198
+ saveFile.click(write_file, inputs=[fileName, fileContent], outputs=completedMessage)
199
+ deleteFile.click(delete_file, inputs=fileName, outputs=completedMessage)
200
+ appendFile.click(append_file, inputs=[fileName, fileContent], outputs=completedMessage )
201
+
202
+
203
+ inputs.submit(predict, [inputs, top_p, temperature,chat_counter, chatbot, state], [chatbot, state, chat_counter])
204
+ b1.click(predict, [inputs, top_p, temperature, chat_counter, chatbot, state], [chatbot, state, chat_counter])
205
+ b1.click(reset_textbox, [], [inputs])
206
+ inputs.submit(reset_textbox, [], [inputs])
207
+ gr.Markdown(description)
208
+
209
+ demo.queue().launch(debug=True)
requirements (7).txt ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ transformers
2
+ torch
3
+ Werkzeug
4
+ huggingface_hub
5
+ Pillow
6
+ datasets