Erikaww commited on
Commit
ca9d207
·
verified ·
1 Parent(s): c32311d

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +25 -27
app.py CHANGED
@@ -1,48 +1,51 @@
1
  import gradio as gr
2
  from huggingface_hub import InferenceClient
3
-
4
 
5
  def respond(
6
- message,
7
- history: list[dict[str, str]],
8
- system_message,
9
- max_tokens,
10
- temperature,
11
- top_p,
12
  hf_token: gr.OAuthToken,
13
  ):
14
  """
15
- 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
 
16
  """
17
- #client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b")
18
- client = InferenceClient(token=hf_token.token, model="apple/FastVLM-7B")
19
- messages = [{"role": "system", "content": system_message}]
 
 
20
 
 
 
21
  messages.extend(history)
22
-
23
  messages.append({"role": "user", "content": message})
24
 
25
  response = ""
26
 
27
- for message in client.chat_completion(
28
- messages,
 
29
  max_tokens=max_tokens,
30
  stream=True,
31
  temperature=temperature,
32
  top_p=top_p,
33
  ):
34
- choices = message.choices
35
  token = ""
36
- if len(choices) and choices[0].delta.content:
37
  token = choices[0].delta.content
38
 
39
  response += token
40
  yield response
41
 
42
 
43
- """
44
- For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
45
- """
46
  chatbot = gr.ChatInterface(
47
  respond,
48
  type="messages",
@@ -50,21 +53,16 @@ chatbot = gr.ChatInterface(
50
  gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
51
  gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
52
  gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
53
- gr.Slider(
54
- minimum=0.1,
55
- maximum=1.0,
56
- value=0.95,
57
- step=0.05,
58
- label="Top-p (nucleus sampling)",
59
- ),
60
  ],
61
  )
62
 
 
63
  with gr.Blocks() as demo:
64
  with gr.Sidebar():
65
  gr.LoginButton()
66
  chatbot.render()
67
 
68
-
69
  if __name__ == "__main__":
70
  demo.launch()
 
 
1
  import gradio as gr
2
  from huggingface_hub import InferenceClient
3
+ from typing import List, Dict
4
 
5
  def respond(
6
+ message: str,
7
+ history: List[Dict[str, str]],
8
+ system_message: str,
9
+ max_tokens: int,
10
+ temperature: float,
11
+ top_p: float,
12
  hf_token: gr.OAuthToken,
13
  ):
14
  """
15
+ Para mais informações sobre o Inference API:
16
+ https://huggingface.co/docs/huggingface_hub/guides/inference
17
  """
18
+ # Inicializa cliente de inferência
19
+ client = InferenceClient(
20
+ token=hf_token.token,
21
+ model="apple/FastVLM-7B"
22
+ )
23
 
24
+ # Prepara mensagens
25
+ messages = [{"role": "system", "content": system_message}]
26
  messages.extend(history)
 
27
  messages.append({"role": "user", "content": message})
28
 
29
  response = ""
30
 
31
+ # Stream de tokens
32
+ for chunk in client.chat_completion(
33
+ messages=messages,
34
  max_tokens=max_tokens,
35
  stream=True,
36
  temperature=temperature,
37
  top_p=top_p,
38
  ):
39
+ choices = chunk.choices
40
  token = ""
41
+ if len(choices) and choices[0].delta and choices[0].delta.content:
42
  token = choices[0].delta.content
43
 
44
  response += token
45
  yield response
46
 
47
 
48
+ # Interface do chatbot
 
 
49
  chatbot = gr.ChatInterface(
50
  respond,
51
  type="messages",
 
53
  gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
54
  gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
55
  gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
56
+ gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
 
 
 
 
 
 
57
  ],
58
  )
59
 
60
+ # Monta layout com Sidebar
61
  with gr.Blocks() as demo:
62
  with gr.Sidebar():
63
  gr.LoginButton()
64
  chatbot.render()
65
 
 
66
  if __name__ == "__main__":
67
  demo.launch()
68
+