Error when running example.py about llama-adapter

[aojunzz]

@reddiamond1234 we don't support the windows platform now.

[randerzander]

This problem isn't limited to Windows. On Ubuntu w/ a CUDA 12 driver:

(llama_adapter) dev@desktop:~/projects/LLaMA-Adapter$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 22.04.2 LTS
Release:        22.04
Codename:       jammy

(llama_adapter) dev@desktop:~/projects/LLaMA-Adapter$ torchrun --nproc_per_node 1 example.py \
         --ckpt_dir $TARGET_FOLDER/model_size\
         --tokenizer_path $TARGET_FOLDER/tokenizer.model \
         --adapter_path $ADAPTER_PATH
Traceback (most recent call last):
  File "example.py", line 114, in <module>                                                                                                                            
    fire.Fire(main)                                                                
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/fire/core.py", line 141, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/fire/core.py", line 475, in _Fire
    component, remaining_args = _CallAndUpdateTrace(
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/fire/core.py", line 691, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "example.py", line 88, in main
    local_rank, world_size = setup_model_parallel()
  File "example.py", line 35, in setup_model_parallel
    torch.distributed.init_process_group("nccl")
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/torch/distributed/distributed_c10d.py", line 907, in init_process_group
    default_pg = _new_process_group_helper(
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/torch/distributed/distributed_c10d.py", line 1013, in _new_process_group_helper
    raise RuntimeError("Distributed package doesn't have NCCL " "built in")
RuntimeError: Distributed package doesn't have NCCL built in
ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: 1) local_rank: 0 (pid: 2214009) of binary: /home/dev/miniconda/envs/llama_adapter/bin/python
Traceback (most recent call last):
  File "/home/dev/miniconda/envs/llama_adapter/bin/torchrun", line 33, in <module>
    sys.exit(load_entry_point('torch==2.0.1', 'console_scripts', 'torchrun')())
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
    return f(*args, **kwargs)
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/torch/distributed/run.py", line 794, in main
    run(args)
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/torch/distributed/run.py", line 785, in run
    elastic_launch(
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
    return launch_agent(self._config, self._entrypoint, list(args))
  File "/home/dev/miniconda/envs/llama_adapter/lib/python3.8/site-packages/torch/distributed/launcher/api.py", line 250, in launch_agent
    raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
============================================================
example.py FAILED
------------------------------------------------------------
Failures:
  <NO_OTHER_FAILURES>
------------------------------------------------------------
Root Cause (first observed failure):
[0]:
  time      : 2023-05-12_11:56:32
  host      : desktop
  rank      : 0 (local_rank: 0)
  exitcode  : 1 (pid: 2214009)
  error_file: <N/A>
  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
============================================================

[randerzander]

In case it helps others, I worked around this problem by:

1. Recreating the llama_adapter conda env
2. first installing the appropriate torch build for my machine (install command picked from here)
3. Remove the torch entry from requirements.txt
4. pip install -r requirements.txt