Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
from fastapi import FastAPI, HTTPException, Request
|
|
|
2 |
from llama_cpp import Llama
|
3 |
import time
|
4 |
import os
|
@@ -7,11 +8,20 @@ app = FastAPI()
|
|
7 |
|
8 |
llm = Llama(model_path="llama-2-7b-chat.Q3_K_S.gguf", n_ctx=2048, n_batch=512, use_mlock=True, n_threads=8)
|
9 |
|
10 |
-
def
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
|
16 |
@app.post("/chat")
|
17 |
async def chat(request: Request):
|
@@ -22,33 +32,18 @@ async def chat(request: Request):
|
|
22 |
if not user_input:
|
23 |
raise HTTPException(status_code=400, detail="Missing 'user_input' field in the request JSON.")
|
24 |
|
25 |
-
os.system("cls")
|
26 |
-
print("Chatbot by Aritra Roy & DVLH")
|
27 |
-
|
28 |
ask = user_input
|
29 |
|
30 |
prompt = f"Llama-2-Chat [INST] <<SYS>>You're an assistant named Tusti. You are Developed by Aritra Roy. Don't share any false information.<</SYS>> {ask} [/INST]"
|
31 |
|
32 |
-
output_stream = llm(prompt, max_tokens=1024, echo=False, temperature=0.2, top_p=0.1, stream=True)
|
33 |
-
|
34 |
-
# Remove the response_text variable and directly print the chunks
|
35 |
-
while True:
|
36 |
-
try:
|
37 |
-
chunk = next(output_stream)
|
38 |
-
if chunk.get('choices') and chunk['choices'][0].get('text'):
|
39 |
-
text = chunk['choices'][0]['text']
|
40 |
-
typewrite(text, delay=0.00)
|
41 |
-
except StopIteration:
|
42 |
-
break
|
43 |
-
|
44 |
if ask == 'clear':
|
45 |
-
|
46 |
|
47 |
-
return
|
48 |
|
49 |
except Exception as e:
|
50 |
raise HTTPException(status_code=500, detail=str(e))
|
51 |
|
52 |
if __name__ == "__main__":
|
53 |
import uvicorn
|
54 |
-
uvicorn.run(app, host="0.0.0.0", port=7860)
|
|
|
1 |
from fastapi import FastAPI, HTTPException, Request
|
2 |
+
from fastapi.responses import StreamingResponse
|
3 |
from llama_cpp import Llama
|
4 |
import time
|
5 |
import os
|
|
|
8 |
|
9 |
llm = Llama(model_path="llama-2-7b-chat.Q3_K_S.gguf", n_ctx=2048, n_batch=512, use_mlock=True, n_threads=8)
|
10 |
|
11 |
+
def generate_output_stream(prompt):
|
12 |
+
output_stream = llm(prompt, max_tokens=1024, echo=False, temperature=0.2, top_p=0.1, stream=True)
|
13 |
+
while True:
|
14 |
+
try:
|
15 |
+
chunk = next(output_stream)
|
16 |
+
if chunk.get('choices') and chunk['choices'][0].get('text'):
|
17 |
+
text = chunk['choices'][0]['text']
|
18 |
+
yield text.encode('utf-8')
|
19 |
+
time.sleep(0.01) # Adjust the delay if needed
|
20 |
+
except StopIteration:
|
21 |
+
break
|
22 |
+
|
23 |
+
def clear_screen():
|
24 |
+
os.system("cls")
|
25 |
|
26 |
@app.post("/chat")
|
27 |
async def chat(request: Request):
|
|
|
32 |
if not user_input:
|
33 |
raise HTTPException(status_code=400, detail="Missing 'user_input' field in the request JSON.")
|
34 |
|
|
|
|
|
|
|
35 |
ask = user_input
|
36 |
|
37 |
prompt = f"Llama-2-Chat [INST] <<SYS>>You're an assistant named Tusti. You are Developed by Aritra Roy. Don't share any false information.<</SYS>> {ask} [/INST]"
|
38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
if ask == 'clear':
|
40 |
+
clear_screen()
|
41 |
|
42 |
+
return StreamingResponse(generate_output_stream(prompt), media_type="text/plain")
|
43 |
|
44 |
except Exception as e:
|
45 |
raise HTTPException(status_code=500, detail=str(e))
|
46 |
|
47 |
if __name__ == "__main__":
|
48 |
import uvicorn
|
49 |
+
uvicorn.run(app, host="0.0.0.0", port=7860)
|