fabianzarncke commited on
Commit
628ed2d
·
verified ·
1 Parent(s): 0ebc8c6

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +85 -0
app.py CHANGED
@@ -1,6 +1,90 @@
1
  import gradio as gr
2
  from huggingface_hub import InferenceClient
3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  """
5
  For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
6
  """
@@ -62,3 +146,4 @@ demo = gr.ChatInterface(
62
 
63
  if __name__ == "__main__":
64
  demo.launch()
 
 
1
  import gradio as gr
2
  from huggingface_hub import InferenceClient
3
 
4
+ # Initialisierung des Inference-Clients
5
+ client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
6
+
7
+
8
+ # Funktion, die die Chat-Antwort generiert
9
+ def respond(
10
+ message,
11
+ history: list[tuple[str, str]],
12
+ system_message,
13
+ max_tokens,
14
+ temperature,
15
+ top_p,
16
+ ):
17
+ messages = [{"role": "system", "content": system_message}]
18
+
19
+ # Historie hinzufügen
20
+ for val in history:
21
+ if val[0]:
22
+ messages.append({"role": "user", "content": val[0]})
23
+ if val[1]:
24
+ messages.append({"role": "assistant", "content": val[1]})
25
+
26
+ # Benutzer-Nachricht hinzufügen
27
+ messages.append({"role": "user", "content": message})
28
+
29
+ response = ""
30
+
31
+ # Stream-Antwort verarbeiten
32
+ for message in client.chat_completion(
33
+ messages,
34
+ max_tokens=max_tokens,
35
+ stream=True,
36
+ temperature=temperature,
37
+ top_p=top_p,
38
+ ):
39
+ token = message.choices[0].delta.content
40
+ response += token
41
+ yield response
42
+
43
+
44
+ # Funktion zur Bearbeitung des Prompts vor dem Senden
45
+ def edit_prompt(original_prompt):
46
+ # Hier könnte man Logik einfügen, um den Prompt automatisch anzupassen
47
+ return original_prompt + "and also tell me what a car looks like." # Zurzeit unverändert zurückgeben
48
+
49
+
50
+ # Gradio UI anpassen
51
+ def chat_interface_wrapper(
52
+ message,
53
+ history,
54
+ system_message,
55
+ max_tokens,
56
+ temperature,
57
+ top_p,
58
+ ):
59
+ # Möglichkeit, den Prompt vor dem Senden zu bearbeiten
60
+ message = edit_prompt(message)
61
+ return respond(message, history, system_message, max_tokens, temperature, top_p)
62
+
63
+
64
+ # Gradio Chat Interface mit zusätzlichen Eingaben
65
+ demo = gr.ChatInterface(
66
+ chat_interface_wrapper,
67
+ additional_inputs=[
68
+ gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
69
+ gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
70
+ gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
71
+ gr.Slider(
72
+ minimum=0.1,
73
+ maximum=1.0,
74
+ value=0.95,
75
+ step=0.05,
76
+ label="Top-p (nucleus sampling)",
77
+ ),
78
+ ],
79
+ )
80
+
81
+ if __name__ == "__main__":
82
+ demo.launch()
83
+
84
+ '''
85
+ import gradio as gr
86
+ from huggingface_hub import InferenceClient
87
+
88
  """
89
  For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
90
  """
 
146
 
147
  if __name__ == "__main__":
148
  demo.launch()
149
+ '''