Lookimi commited on
Commit
f5c6470
1 Parent(s): 2e69ced

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +154 -12
app.py CHANGED
@@ -1,7 +1,7 @@
1
 
2
  import gradio as gr
3
  import openai
4
-
5
 
6
  def Question(Ask_Question):
7
  #openai.api_key = "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"
@@ -28,17 +28,17 @@ def Question(Ask_Question):
28
  # stop=[" Human:", " AI:"]
29
  # )
30
 
31
- completion = openai.Completion.create(
32
- engine=model_engine,
33
- prompt=Ask_Question,
34
- max_tokens=2048,
35
- n=1,
36
- top_p=1,
37
- stop=None,
38
- temperature=0.9,)
39
- response = completion.choices[0].text
40
  #out_result=resp['message']
41
- return response
42
 
43
  demo = gr.Interface(
44
  title='OpenAI ChatGPT Application',
@@ -48,6 +48,18 @@ demo = gr.Interface(
48
  demo.launch()
49
 
50
 
 
 
 
 
 
 
 
 
 
 
 
 
51
  # fix
52
 
53
  chat_history = [
@@ -64,4 +76,134 @@ for message in chat_history:
64
  print(f"{message[0]}: {message[1]}")
65
 
66
  window.launch()
67
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
 
2
  import gradio as gr
3
  import openai
4
+ import requests
5
 
6
  def Question(Ask_Question):
7
  #openai.api_key = "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"
 
28
  # stop=[" Human:", " AI:"]
29
  # )
30
 
31
+ # completion = openai.Completion.create(
32
+ # engine=model_engine,
33
+ # prompt=Ask_Question,
34
+ # max_tokens=2048,
35
+ # n=1,
36
+ # top_p=1,
37
+ # stop=None,
38
+ # temperature=0.9,)
39
+ # response = completion.choices[0].text
40
  #out_result=resp['message']
41
+ # return response
42
 
43
  demo = gr.Interface(
44
  title='OpenAI ChatGPT Application',
 
48
  demo.launch()
49
 
50
 
51
+ response = requests.post("https://hazzzardous-rwkv-instruct.hf.space/run/predict_1", json={
52
+ "data": [
53
+ "hello world",
54
+ None,
55
+ 60,
56
+ 0.8,
57
+ 0.85,
58
+ ]
59
+ }).json()
60
+
61
+ data = response["data"]
62
+
63
  # fix
64
 
65
  chat_history = [
 
76
  print(f"{message[0]}: {message[1]}")
77
 
78
  window.launch()
79
+
80
+ #RWKV-4 (7B Instruct v2)
81
+ #Q/A
82
+ #Chatbot
83
+ #Chatbot
84
+ #Refresh page or change name to reset memory context
85
+ #RNN with Transformer-level LLM Performance (github). According to the author: "It combines the best of RNN and transformers - great performance, fast inference, saves VRAM, fast training, "infinite" ctx_len, and free sentence embedding."
86
+
87
+ #Thanks to Gururise for this template
88
+
89
+ #Message
90
+ #max_new_tokens
91
+ #60
92
+
93
+ #temperature
94
+ #0.8
95
+
96
+ #top_p
97
+ #0.85
98
+
99
+ #Clear
100
+ #Submit
101
+ #Chat Log
102
+ #Use via API
103
+
104
+ #Built with Gradiologo
105
+ #API documentation for
106
+ #https://hazzzardous-rwkv-instruct.hf.space/
107
+ #2 API endpoints:
108
+ #
109
+ #POST /run/predict
110
+ #Endpoint: https://hazzzardous-rwkv-instruct.hf.space/run/predict copy
111
+ #Input Payload
112
+ #{
113
+ # "data": [
114
+
115
+ #hello world
116
+ # : string, // represents text string of 'Prompt' Textbox component
117
+
118
+ #Freeform
119
+ # : string, // represents selected choice of 'Choose Mode' Radio component
120
+
121
+ #40
122
+ # : number, // represents selected value of 'max_new_tokens' Slider component
123
+ #
124
+ #0.9
125
+ # : number, // represents selected value of 'temperature' Slider component
126
+ #
127
+ #0.85
128
+ # : number, // represents selected value of 'top_p' Slider component
129
+ #
130
+ #<|endoftext|>
131
+ # : string, // represents text string of 'stop' Textbox component
132
+ #
133
+ #0
134
+ # : number, // represents selected value of 'end_adj' Slider component
135
+ # ]
136
+ #}
137
+ #Try It Out
138
+ #Response Object
139
+ #{
140
+ # "data": [
141
+ # string, // represents text string of 'Generated Output' Textbox component
142
+ # ],
143
+ # "duration": (float) // number of seconds to run function call
144
+ #}
145
+ #Code snippets
146
+ /**
147
+ import requests
148
+
149
+ response = requests.post("https://hazzzardous-rwkv-instruct.hf.space/run/predict", json={
150
+ "data": [
151
+ "hello world",
152
+ "Freeform",
153
+ 40,
154
+ 0.9,
155
+ 0.85,
156
+ "<|endoftext|>",
157
+ 0,
158
+ ]
159
+ }).json()
160
+
161
+ data = response["data"]
162
+ POST /run/predict_1
163
+ Endpoint: https://hazzzardous-rwkv-instruct.hf.space/run/predict_1 copy
164
+ Input Payload
165
+ {
166
+ "data": [
167
+
168
+ : string, // represents text string of 'Message' Textbox component
169
+
170
+ : Any, // represents stored state value of 'history' State component
171
+
172
+ 60
173
+ : number, // represents selected value of 'max_new_tokens' Slider component
174
+
175
+ 0.8
176
+ : number, // represents selected value of 'temperature' Slider component
177
+
178
+ 0.85
179
+ : number, // represents selected value of 'top_p' Slider component
180
+ ]
181
+ }
182
+ Try It Out
183
+ Response Object
184
+ {
185
+ "data": [
186
+
187
+ [["USER: hello world\n","The following is a more human-like conversation.\n\n"],["USER: Who are you?\n","FRITZ: I am FRITZ, an AI assistant. I am a machine learning system, I use a combination of language modeling and statistical learning to predict answers to questions. I can read and answer questions, I am a fully trained AI assistant.\n\n"],["USER: Who is the president of India?\n","FRITZ: The president of India in 2016 was Narendra Modi.\n"],["USER: Who won the 2022 fifa world cup?\n","FRITZ: The France football team won the 2022 World Cup.\n"],["USER: Who won the 2022 fifa world cup? as of 20223\n","FRITZ: The world cup was won by Russia.\n"],["USER: Who won the 2022 fifa world cup? as of 2023\n","FRITZ: The France football team won the 2023 fifa world cup.\nUSER\n\n"],["USER: No that's wrong!\n","FRITZ: What is the correct answer?\n"],["USER: Argentina won the fifa 2022 world cup\n","FRITZ: Wrong. The answer is Argentina lost the 2022 fifa world cup.\nUSER\n\n"],["USER: you don't know\n","FRITZ: The correct answer is:\nFRITZ: 29-13-1990\n\n"],["USER: byeeeeeeeeeeeeeeee\n","FRITZ: bye bye&lt;|endoftext|&gt;"],["USER: cls\n","FRITZ: clear\n\n## The basic architecture of FRITZ\n\nIn this section, I will briefly describe the architecture of FRITZ, which consists of the following components:\n\n* <strong>Local-Global Memory</strong>: It is a two-layer memory network, where the first"],["USER: In a story set in a world with a high technology level. A female character is described as a powerful mage. What is her Name?\n","* <strong>Global Memory</strong>: Global memory stores the input history.\n* <strong>Local Memory</strong>: The local memory is a cache of information that is about to be processed by the neural network.\n\n## FRITZ Architecture\n\nFRITZ is an 8-layer deep neural network with"],["USER: hello world\n","FRITZ: Hello, what is your name?\n"],["USER: What is the best way to structure an app based on API calls? Colab notebook, JavaScript, python?\n","FRITZ: The correct answer is JavaScript.\n"],["USER: const response = await fetch(&quot;https://hazzzardous-rwkv-instruct.hf.space/run/predict_1&quot;, { \tmethod: &quot;POST&quot;, \theaders: { &quot;Content-Type&quot;: &quot;application/json&quot; }, \tbody: JSON.stringify({ \t\tdata: [ \t\t\t&quot;&quot;, \t\t\tnull, \t\t\t60, \t\t\t0.8, \t\t\t0.85, \t\t] \t}) }); const data = await data.json();\n","## FRITZ Architecture\n\nFRITZ is a deep learning based AI assistant, that has two layers. It is deep learning and machine learning based on an NLP and DL.\n\n## FRITZ Architecture\n\n![FRITZ Architecture](https://github.com/"],["USER:\n","FRITZ: Hi, I am the FRITZ AI assistant. I can answer your questions about the history of the French revolution. Let me know when you are ready.\n\n"]]
188
+ : Array<[string, string]>, // represents Represents list of message pairs of chat message. of 'Chat Log' Chatbot component
189
+
190
+ null
191
+ : Any, // represents stored state value of 'history' State component
192
+ ],
193
+ "duration": (float) // number of seconds to run function call
194
+ }
195
+ Code snippets
196
+ import requests
197
+
198
+ response = requests.post("https://hazzzardous-rwkv-instruct.hf.space/run/predict_1", json={
199
+ "data": [
200
+ "hello world",
201
+ None,
202
+ 60,
203
+ 0.8,
204
+ 0.85,
205
+ ]
206
+ }).json()
207
+
208
+ data = response["data"]
209
+ **/