Cannot use /v1/embeddings api with BGE-M3 model

fastchat version 0.2.36

First, start the services:

python3 -m fastchat.serve.controller --port 21002 --host 0.0.0.0
python3 -m fastchat.serve.model_worker --port 21003 --model-names bge-m3 --model-path /root/proj/models/bge-m3 --device cpu --controller-address http://0.0.0.0:21002 --host 0.0.0.0
python3 -m fastchat.serve.openai_api_server --controller-address http://0.0.0.0:21002 --port 8001 --host 0.0.0.0

Then, post requests to the 8001 port:

import requests

url = "http://10.99.106.214:8001/v1/embeddings"
headers = {
"Content-Type": "application/json"
}
data = {
"model": "bge-m3",
"input": "Hello world!"
}

response = requests.post(url, headers=headers, json=data)

print(response.text)_

and get the response:
{"object":"error","message":"Not Found","code":50001}

Reproduce the reported setup with the controller, model_worker, and openai_api_server commands using the bge-m3 model, then inspect how the /v1/embeddings endpoint routes requests. Done means the posted request no longer returns the Not Found error and produces an embedding response; no test or source file is named in the issue.