Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -4,7 +4,7 @@ import logging
|
|
4 |
from fastapi import FastAPI, HTTPException
|
5 |
from fastapi.responses import JSONResponse
|
6 |
from pydantic import BaseModel
|
7 |
-
from huggingface_hub import InferenceClient
|
8 |
from typing import Optional
|
9 |
|
10 |
# Set up logging
|
@@ -25,25 +25,33 @@ class ChatResponse(BaseModel):
|
|
25 |
response: str
|
26 |
status: str
|
27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
def llm_chat_response(text: str) -> str:
|
29 |
try:
|
30 |
-
|
31 |
-
logger.info("Checking HF_TOKEN...")
|
32 |
-
if not HF_TOKEN:
|
33 |
-
logger.error("HF_TOKEN not found in environment variables")
|
34 |
-
raise HTTPException(status_code=500, detail="HF_TOKEN not configured")
|
35 |
|
36 |
-
|
37 |
-
client = InferenceClient(token=HF_TOKEN)
|
38 |
-
|
39 |
-
logger.info("Sending request to model...")
|
40 |
-
# Using text-generation endpoint instead of chat
|
41 |
response = client.text_generation(
|
42 |
-
text +
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
)
|
|
|
|
|
47 |
return response
|
48 |
|
49 |
except Exception as e:
|
|
|
4 |
from fastapi import FastAPI, HTTPException
|
5 |
from fastapi.responses import JSONResponse
|
6 |
from pydantic import BaseModel
|
7 |
+
from huggingface_hub import InferenceClient, HfApi
|
8 |
from typing import Optional
|
9 |
|
10 |
# Set up logging
|
|
|
25 |
response: str
|
26 |
status: str
|
27 |
|
28 |
+
# Initialize HF client at startup
|
29 |
+
try:
|
30 |
+
HF_TOKEN = os.getenv("HF_TOKEN")
|
31 |
+
if not HF_TOKEN:
|
32 |
+
logger.warning("HF_TOKEN not found in environment variables")
|
33 |
+
api = HfApi(token=HF_TOKEN)
|
34 |
+
client = InferenceClient(
|
35 |
+
model="meta-llama/Llama-3.2-11B-Vision-Instruct", # You might need to change this
|
36 |
+
token=HF_TOKEN
|
37 |
+
)
|
38 |
+
logger.info("Successfully initialized HuggingFace client")
|
39 |
+
except Exception as e:
|
40 |
+
logger.error(f"Error initializing HuggingFace client: {str(e)}")
|
41 |
+
|
42 |
def llm_chat_response(text: str) -> str:
|
43 |
try:
|
44 |
+
logger.info(f"Processing text: {text}")
|
|
|
|
|
|
|
|
|
45 |
|
46 |
+
# Direct text generation
|
|
|
|
|
|
|
|
|
47 |
response = client.text_generation(
|
48 |
+
text + " describe in one line only",
|
49 |
+
max_new_tokens=100,
|
50 |
+
temperature=0.7,
|
51 |
+
repetition_penalty=1.2
|
52 |
)
|
53 |
+
|
54 |
+
logger.info(f"Generated response: {response}")
|
55 |
return response
|
56 |
|
57 |
except Exception as e:
|