Local LangChain with FastChat Error

[code-ex0]

I want to self how open ai API with fastchat and I see this example but i see a error.

I can use the seconde command python3 -m fastchat.serve.model_worker --model-names "gpt-3.5-turbo,text-davinci-003,text-embedding-ada-002" --model-path /path/to/weights

the parameter --model-names doesn't exist and I don't know how to replace it.

 python -m fastchat.serve.model_worker --model-names "gpt-3.5-turbo,text-davinci-003,text-embedding-ada-002" --model-path lmsys/fastchat-t5-3b-v1.0
2023-06-13 15:07:33 | ERROR | stderr | usage: model_worker.py [-h] [--host HOST] [--port PORT] [--worker-address WORKER_ADDRESS] [--controller-address CONTROLLER_ADDRESS] [--model-path MODEL_PATH] [--device {cpu,cuda,m
ps}] [--gpus GPUS]
2023-06-13 15:07:33 | ERROR | stderr |                        [--num-gpus NUM_GPUS] [--max-gpu-memory MAX_GPU_MEMORY] [--load-8bit] [--cpu-offloading] [--model-name MODEL_NAME] [--limit-model-concurrency LIMIT_MODEL_CO
NCURRENCY]
2023-06-13 15:07:33 | ERROR | stderr |                        [--stream-interval STREAM_INTERVAL] [--no-register]
2023-06-13 15:07:33 | ERROR | stderr | model_worker.py: error: unrecognized arguments: --model-names gpt-3.5-turbo,text-davinci-003,text-embedding-ada-002

do you have an idea ??

[IlianP]

It seems you forgot about this part:

Replace /path/to/weights below with the a real path to a local model such as Vicuna. It can also be a Hugging Face repo id such as lmsys/fastchat-t5-3b-v1.0.

[code-ex0]

i use this python -m fastchat.serve.model_worker --model-names "gpt-3.5-turbo,text-davinci-003,text-embedding-ada-002" --model-path lmsys/fastchat-t5-3b-v1.0 i dont forgot the model path

[creatorone1]

mybe you should check your FastChat/fastchat/serve/model_worker.pyfile. serch "--model-names" in this file?

[volksen]

the parameter is called --model-name. this is a small bug in the example. you can check the parameters with python3 -m fastchat.serve.model_worker --help

[code-ex0]

Ok I find error if I install fschat 0.2.13 he solves the problems of the doc. pip install fschat==0.2.13