Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from huggingface_hub import InferenceClient
|
2 |
+
from fastapi import FastAPI, Request
|
3 |
+
from pydantic import BaseModel
|
4 |
+
import uvicorn
|
5 |
+
import os
|
6 |
+
|
7 |
+
app = FastAPI()
|
8 |
+
|
9 |
+
MODEL = "mistralai/Mistral-7B-Instruct-v0.1"
|
10 |
+
HF_TOKEN = os.environ["HF_TOKEN"]
|
11 |
+
|
12 |
+
client = InferenceClient(model=MODEL, token=HF_TOKEN)
|
13 |
+
|
14 |
+
class Prompt(BaseModel):
|
15 |
+
message: str
|
16 |
+
|
17 |
+
@app.post("/chat")
|
18 |
+
async def chat(prompt: Prompt):
|
19 |
+
system_prompt = (
|
20 |
+
"You are a beginner programming student helping a peer. "
|
21 |
+
"Offer hints, ask questions, and support understanding—don’t give full solutions."
|
22 |
+
)
|
23 |
+
full_prompt = f"<s>[INST] <<SYS>>{system_prompt}<</SYS>>\n{prompt.message} [/INST]"
|
24 |
+
|
25 |
+
output = client.text_generation(
|
26 |
+
prompt=full_prompt,
|
27 |
+
max_new_tokens=200,
|
28 |
+
temperature=0.7,
|
29 |
+
top_p=0.95,
|
30 |
+
do_sample=True
|
31 |
+
)
|
32 |
+
return {"reply": output.strip()}
|