Spaces:
Running
Running
nguyenbh
commited on
Commit
·
7fc85f7
1
Parent(s):
692b3f2
Update app
Browse files
app.py
CHANGED
|
@@ -1,20 +1,17 @@
|
|
| 1 |
import os
|
| 2 |
import gradio as gr
|
| 3 |
from azure.ai.inference import ChatCompletionsClient
|
|
|
|
| 4 |
from azure.core.credentials import AzureKeyCredential
|
| 5 |
|
| 6 |
|
| 7 |
# Azure Inference setup
|
| 8 |
url = os.getenv("Azure_Endpoint")
|
| 9 |
api_key = AzureKeyCredential(os.getenv("Azure_API_KEY"))
|
| 10 |
-
|
| 11 |
|
| 12 |
# Initialize the ChatCompletionsClient
|
| 13 |
-
client = ChatCompletionsClient(
|
| 14 |
-
endpoint=url,
|
| 15 |
-
credential=api_key,
|
| 16 |
-
stream=True
|
| 17 |
-
)
|
| 18 |
|
| 19 |
# Get and print model information (optional)
|
| 20 |
try:
|
|
@@ -46,32 +43,35 @@ def get_azure_response(message, chat_history, temperature, max_tokens, top_p, pr
|
|
| 46 |
Function to get a response from the Azure Phi-4 model
|
| 47 |
"""
|
| 48 |
# Prepare conversation history in the format expected by Azure
|
| 49 |
-
messages
|
| 50 |
|
| 51 |
# Add conversation history
|
| 52 |
for human, assistant in chat_history:
|
| 53 |
-
messages.append(
|
| 54 |
-
if assistant:
|
| 55 |
-
messages.append(
|
| 56 |
|
| 57 |
# Add the current message
|
| 58 |
-
messages.append(
|
| 59 |
-
|
| 60 |
-
# Prepare the
|
| 61 |
-
|
| 62 |
-
"messages": messages,
|
| 63 |
"max_tokens": max_tokens,
|
| 64 |
"temperature": temperature,
|
| 65 |
"top_p": top_p,
|
| 66 |
-
"presence_penalty":
|
| 67 |
-
"frequency_penalty":
|
| 68 |
-
"stream": True
|
| 69 |
}
|
| 70 |
|
| 71 |
# Get response
|
| 72 |
try:
|
| 73 |
print("Sending request to Azure...")
|
| 74 |
-
response = client.complete(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
return response
|
| 76 |
except Exception as e:
|
| 77 |
print(f"Error getting response: {str(e)}")
|
|
|
|
| 1 |
import os
|
| 2 |
import gradio as gr
|
| 3 |
from azure.ai.inference import ChatCompletionsClient
|
| 4 |
+
from azure.ai.inference.models import SystemMessage, UserMessage, AssistantMessage
|
| 5 |
from azure.core.credentials import AzureKeyCredential
|
| 6 |
|
| 7 |
|
| 8 |
# Azure Inference setup
|
| 9 |
url = os.getenv("Azure_Endpoint")
|
| 10 |
api_key = AzureKeyCredential(os.getenv("Azure_API_KEY"))
|
| 11 |
+
model_name = os.getenv("Azure_Model_Name")
|
| 12 |
|
| 13 |
# Initialize the ChatCompletionsClient
|
| 14 |
+
client = ChatCompletionsClient(endpoint=url, credential=api_key, stream=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
# Get and print model information (optional)
|
| 17 |
try:
|
|
|
|
| 43 |
Function to get a response from the Azure Phi-4 model
|
| 44 |
"""
|
| 45 |
# Prepare conversation history in the format expected by Azure
|
| 46 |
+
messages=[SystemMessage("You are a helpful AI assistant.")]
|
| 47 |
|
| 48 |
# Add conversation history
|
| 49 |
for human, assistant in chat_history:
|
| 50 |
+
messages.append(UserMessage(human))
|
| 51 |
+
if assistant:
|
| 52 |
+
messages.append(AssistantMessage(assistant))
|
| 53 |
|
| 54 |
# Add the current message
|
| 55 |
+
messages.append(UserMessage(message))
|
| 56 |
+
|
| 57 |
+
# Prepare the model_extras
|
| 58 |
+
model_extras = {
|
|
|
|
| 59 |
"max_tokens": max_tokens,
|
| 60 |
"temperature": temperature,
|
| 61 |
"top_p": top_p,
|
| 62 |
+
"presence_penalty": 0,
|
| 63 |
+
"frequency_penalty": 0,
|
|
|
|
| 64 |
}
|
| 65 |
|
| 66 |
# Get response
|
| 67 |
try:
|
| 68 |
print("Sending request to Azure...")
|
| 69 |
+
response = client.complete(
|
| 70 |
+
model=model_name,
|
| 71 |
+
messages=messages,
|
| 72 |
+
stream=True,
|
| 73 |
+
model_extras=model_extras
|
| 74 |
+
)
|
| 75 |
return response
|
| 76 |
except Exception as e:
|
| 77 |
print(f"Error getting response: {str(e)}")
|