asthaa30 commited on
Commit
6cb707e
·
verified ·
1 Parent(s): 04fbf66

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +28 -16
app.py CHANGED
@@ -11,7 +11,6 @@ MODEL = "nomiChroma3.1"
11
 
12
  client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
13
 
14
-
15
  def respond(
16
  message,
17
  history: list[tuple[str, str]],
@@ -21,28 +20,41 @@ def respond(
21
  top_p,
22
  ):
23
  messages = [{"role": "system", "content": system_message}]
24
-
25
  for val in history:
26
  if val[0]:
27
  messages.append({"role": "user", "content": val[0]})
28
  if val[1]:
29
  messages.append({"role": "assistant", "content": val[1]})
30
-
31
  messages.append({"role": "user", "content": message})
32
-
33
  response = ""
34
-
35
- for message in client.chat_completion(
36
- messages,
37
- max_tokens=max_tokens,
38
- stream=True,
39
- temperature=temperature,
40
- top_p=top_p,
41
- ):
42
- token = message.choices[0].delta.content
43
-
44
- response += token
45
- yield response
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
46
 
47
  """
48
  For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
 
11
 
12
  client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
13
 
 
14
  def respond(
15
  message,
16
  history: list[tuple[str, str]],
 
20
  top_p,
21
  ):
22
  messages = [{"role": "system", "content": system_message}]
 
23
  for val in history:
24
  if val[0]:
25
  messages.append({"role": "user", "content": val[0]})
26
  if val[1]:
27
  messages.append({"role": "assistant", "content": val[1]})
 
28
  messages.append({"role": "user", "content": message})
 
29
  response = ""
30
+ try:
31
+ for message in client.chat_completion(
32
+ messages,
33
+ max_tokens=max_tokens,
34
+ stream=True,
35
+ temperature=temperature,
36
+ top_p=top_p,
37
+ ):
38
+ try:
39
+ # Try to access the content directly
40
+ token = message.choices[0].delta.content
41
+ except AttributeError:
42
+ # If above fails, the response might be in a different format
43
+ # Print the raw message for debugging
44
+ print(f"Debug - Raw message: {message}")
45
+ # Try to parse as JSON
46
+ try:
47
+ parsed_message = json.loads(str(message))
48
+ token = parsed_message.get('choices', [{}])[0].get('delta', {}).get('content', '')
49
+ except json.JSONDecodeError as e:
50
+ print(f"JSON Decode Error: {e}")
51
+ print(f"Problematic message: {message}")
52
+ token = "" # or handle this case as appropriate
53
+ response += token
54
+ yield response
55
+ except Exception as e:
56
+ print(f"An error occurred: {e}")
57
+ yield f"An error occurred: {e}"
58
 
59
  """
60
  For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface