Spaces:
Paused
Paused
Commit
•
86f94f0
1
Parent(s):
0d521c3
Update main.py
Browse files
main.py
CHANGED
|
@@ -34,12 +34,45 @@ async def index():
|
|
| 34 |
class ChatCompletionRequest(BaseModel):
|
| 35 |
prompt: str
|
| 36 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
@app.post("/v1/chat/completions")
|
| 38 |
async def chat(request: ChatCompletionRequest, response_mode=None):
|
| 39 |
tokens = llm.tokenize(request.prompt)
|
| 40 |
async def server_sent_events(chat_chunks, llm):
|
| 41 |
for token in llm.generate(chat_chunks):
|
| 42 |
yield llm.detokenize(token)
|
|
|
|
| 43 |
|
| 44 |
return EventSourceResponse(server_sent_events(tokens, llm))
|
| 45 |
|
|
|
|
| 34 |
class ChatCompletionRequest(BaseModel):
|
| 35 |
prompt: str
|
| 36 |
|
| 37 |
+
|
| 38 |
+
@app.get("/demo")
|
| 39 |
+
async def demo():
|
| 40 |
+
html_content = """
|
| 41 |
+
<!DOCTYPE html>
|
| 42 |
+
<html>
|
| 43 |
+
<body>
|
| 44 |
+
<div id="content"></div>
|
| 45 |
+
|
| 46 |
+
<script>
|
| 47 |
+
var source = new EventSource("https://matthoffner-starchat-alpha.hf.space/stream");
|
| 48 |
+
source.onmessage = function(event) {
|
| 49 |
+
document.getElementById("content").innerHTML += event.data
|
| 50 |
+
};
|
| 51 |
+
</script>
|
| 52 |
+
|
| 53 |
+
</body>
|
| 54 |
+
</html>
|
| 55 |
+
"""
|
| 56 |
+
return HTMLResponse(content=html_content, status_code=200)
|
| 57 |
+
|
| 58 |
+
@app.get("/stream")
|
| 59 |
+
async def chat(prompt = "Write a simple express erver"):
|
| 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 |
@app.post("/v1/chat/completions")
|
| 70 |
async def chat(request: ChatCompletionRequest, response_mode=None):
|
| 71 |
tokens = llm.tokenize(request.prompt)
|
| 72 |
async def server_sent_events(chat_chunks, llm):
|
| 73 |
for token in llm.generate(chat_chunks):
|
| 74 |
yield llm.detokenize(token)
|
| 75 |
+
yield ""
|
| 76 |
|
| 77 |
return EventSourceResponse(server_sent_events(tokens, llm))
|
| 78 |
|