Spaces:
Runtime error
Runtime error
Commit
·
4381f13
1
Parent(s):
271190b
Delete main.py
Browse files
main.py
DELETED
|
@@ -1,85 +0,0 @@
|
|
| 1 |
-
import fastapi
|
| 2 |
-
import json
|
| 3 |
-
import markdown
|
| 4 |
-
import uvicorn
|
| 5 |
-
from fastapi.responses import HTMLResponse
|
| 6 |
-
from fastapi.middleware.cors import CORSMiddleware
|
| 7 |
-
from sse_starlette.sse import EventSourceResponse
|
| 8 |
-
from ctransformers import AutoModelForCausalLM
|
| 9 |
-
from ctransformers.langchain import CTransformers
|
| 10 |
-
from pydantic import BaseModel
|
| 11 |
-
|
| 12 |
-
llm = AutoModelForCausalLM.from_pretrained("TheBloke/gorilla-7B-GGML",
|
| 13 |
-
model_file="Gorilla-7B.ggmlv3.q4_0.bin",
|
| 14 |
-
model_type="llama")
|
| 15 |
-
app = fastapi.FastAPI()
|
| 16 |
-
app.add_middleware(
|
| 17 |
-
CORSMiddleware,
|
| 18 |
-
allow_origins=["*"],
|
| 19 |
-
allow_credentials=True,
|
| 20 |
-
allow_methods=["*"],
|
| 21 |
-
allow_headers=["*"],
|
| 22 |
-
)
|
| 23 |
-
|
| 24 |
-
@app.get("/")
|
| 25 |
-
async def index():
|
| 26 |
-
html_content = """
|
| 27 |
-
<!DOCTYPE html>
|
| 28 |
-
<html>
|
| 29 |
-
<style>
|
| 30 |
-
body {
|
| 31 |
-
font-family: "Arial";
|
| 32 |
-
}
|
| 33 |
-
h1 {
|
| 34 |
-
text-align: "center";
|
| 35 |
-
}
|
| 36 |
-
</style>
|
| 37 |
-
<body>
|
| 38 |
-
<h1>gorilla</h1>
|
| 39 |
-
<input id="prompt" type="text">
|
| 40 |
-
<button id="search">I'm feeling lucky</button>
|
| 41 |
-
<div id="content"></div>
|
| 42 |
-
<script>
|
| 43 |
-
document.getElementById("search").addEventListener("click", () => {
|
| 44 |
-
let prompt = document.getElementById("prompt").value;
|
| 45 |
-
let source = new EventSource(`https://matthoffner-gorilla.hf.space/stream?prompt=${prompt}`);
|
| 46 |
-
source.onmessage = function(event) {
|
| 47 |
-
console.log(event);
|
| 48 |
-
let eventData = event.data;
|
| 49 |
-
document.getElementById("content").innerHTML += eventData
|
| 50 |
-
};
|
| 51 |
-
});
|
| 52 |
-
</script>
|
| 53 |
-
</body>
|
| 54 |
-
</html>
|
| 55 |
-
"""
|
| 56 |
-
return HTMLResponse(content=html_content, status_code=200)
|
| 57 |
-
|
| 58 |
-
@app.get("/stream")
|
| 59 |
-
async def chat(prompt = "I want to download a dataset from GCS"):
|
| 60 |
-
tokens = llm.tokenize(prompt)
|
| 61 |
-
async def server_sent_events(chat_chunks, llm):
|
| 62 |
-
yield prompt
|
| 63 |
-
for chat_chunk in llm.generate(chat_chunks):
|
| 64 |
-
yield llm.detokenize(chat_chunk)
|
| 65 |
-
yield ""
|
| 66 |
-
|
| 67 |
-
return EventSourceResponse(server_sent_events(tokens, llm))
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
class ChatCompletionRequest(BaseModel):
|
| 71 |
-
messages: str
|
| 72 |
-
|
| 73 |
-
@app.post("/v1/chat/completions")
|
| 74 |
-
async def chat(request: ChatCompletionRequest, response_mode=None):
|
| 75 |
-
tokens = llm.tokenize(request.messages.join(' '))
|
| 76 |
-
async def server_sent_events(chat_chunks, llm):
|
| 77 |
-
for chat_chunk in llm.generate(chat_chunks):
|
| 78 |
-
yield llm.detokenize(chat_chunk)
|
| 79 |
-
yield ""
|
| 80 |
-
|
| 81 |
-
return EventSourceResponse(server_sent_events(tokens, llm))
|
| 82 |
-
|
| 83 |
-
if __name__ == "__main__":
|
| 84 |
-
uvicorn.run(app, host="0.0.0.0", port=8000)
|
| 85 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|