gbx_lm.manage和gbx_lm.fastapi_server入口报错

[anrgct]

模型就在目录下，manage和fastapi_server入口报错

╭─   ~/llm_models/GreenBitAI                                            took  8s  gbai_mlx_lm
╰─❯ python -m gbx_lm.generate --model Qwen-3-30B-A3B-layer-mix-bpw-4.0-mlx  --max-tokens 100 --prompt "calculate 4*8+1024="
==========
<think>
Okay, let's see. The user wants me to calculate 4 times 8 plus 1024. Hmm, I need to remember the order of operations here. So, according to PEMDAS, multiplication comes before addition. That means I should do 4 multiplied by 8 first.

Alright, 4 times 8 is 32. Then I add that result to 1024. So 32 plus 1024. Let
==========
Prompt: 53 tokens, 55.770 tokens-per-sec
Generation: 100 tokens, 83.670 tokens-per-sec
Peak memory: 17.239 GB

╭─   ~/llm_models/GreenBitAI                                           took  15s  gbai_mlx_lm
╰─❯ python -m gbx_lm.manage --scan
Traceback (most recent call last):
  File "/Users/anrgct/miniforge3/envs/gbai_mlx_lm/lib/python3.9/runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/Users/anrgct/miniforge3/envs/gbai_mlx_lm/lib/python3.9/runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/manage.py", line 4, in <module>
    from transformers.commands.user import tabulate
ModuleNotFoundError: No module named 'transformers.commands.user'

╭─   ~/llm_models/GreenBitAI                                                       gbai_mlx_lm
╰─❯ python -m gbx_lm.fastapi_server --model ./Qwen-3-30B-A3B-layer-mix-bpw-4.0-mlx
2025-05-13 20:05:59,343 - gbx_server - INFO - Starting GBX-Model API server. Log file: /Users/anrgct/workspace/gbx-lm/logs/server_20250513_200559.log
2025-05-13 20:05:59,343 - gbx_server - ERROR - Error loading confidence scorers: No module named 'gbx_lm.routing.libra_router.ue_router'
2025-05-13 20:05:59,343 - gbx_server - INFO - Using gbx-lm to load model from ./Qwen-3-30B-A3B-layer-mix-bpw-4.0-mlx
2025-05-13 20:05:59,344 - gbx_server - ERROR - Invalid model path: Qwen-3-30B-A3B-layer-mix-bpw-4.0-mlx
2025-05-13 20:05:59,344 - gbx_server - ERROR - Model path validation failed: Local models must be relative to the current working dir.
2025-05-13 20:05:59,344 - gbx_server - ERROR - Failed to load model from ./Qwen-3-30B-A3B-layer-mix-bpw-4.0-mlx: Local models must be relative to the current working dir.
2025-05-13 20:05:59,344 - gbx_server - ERROR - Failed to initialize ModelProvider: Local models must be relative to the current working dir.
Traceback (most recent call last):
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/fastapi_server.py", line 812, in main
    app, server_config, logger = create_app(args)
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/fastapi_server.py", line 300, in create_app
    model_provider = ModelProvider(server_config.model_config)
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/fastapi_server.py", line 179, in __init__
    self.load(self.cli_args.model)
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/fastapi_server.py", line 221, in load
    self._validate_model_path(model_path)
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/fastapi_server.py", line 189, in _validate_model_path
    raise RuntimeError("Local models must be relative to the current working dir.")
RuntimeError: Local models must be relative to the current working dir.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/anrgct/miniforge3/envs/gbai_mlx_lm/lib/python3.9/runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/Users/anrgct/miniforge3/envs/gbai_mlx_lm/lib/python3.9/runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/fastapi_server.py", line 820, in <module>
    main()
  File "/Users/anrgct/workspace/gbx-lm/gbx_lm/fastapi_server.py", line 816, in main
    logger.error(f"Server startup failed: {str(e)}")
UnboundLocalError: local variable 'logger' referenced before assignment