Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Checkpoint Cannot Be Found For Llama 405B Model #44

Open
dawenxi-007 opened this issue Aug 28, 2024 · 1 comment
Open

Checkpoint Cannot Be Found For Llama 405B Model #44

dawenxi-007 opened this issue Aug 28, 2024 · 1 comment

Comments

@dawenxi-007
Copy link

Trying to run inference with FP8 version of Llama 3.1 405B model (Meta-Llama3.1-405B-Instruct). The model was downloaded with llama download --source huggingface --model-id Meta-Llama3.1-405B-Instruct --hf-token TOKEN. However, the command llama distribution start --name local-llama-405b --port 5000 --disable-ipv6 gave the following error:

------------------------------------------------------------
Failures:
  <NO_OTHER_FAILURES>
------------------------------------------------------------
Root Cause (first observed failure):
[0]:
  time      : 2024-08-25_04:55:10
  host      : node007
  rank      : 0 (local_rank: 0)
  exitcode  : 1 (pid: 3088723)
  error_file: /tmp/torchelastic_kvox4nb5/ee89349c-cc4c-43c4-9796-1ceeb2986a3b_ugr5p160/attempt_0/0/error.json
  traceback : Traceback (most recent call last):
    File "/home/ubuntu/miniforge3/envs/local-llama-405b/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 348, in wrapper
      return f(*args, **kwargs)
    File "/home/ubuntu/taoz/llama-stack/llama_toolchain/inference/meta_reference/parallel_utils.py", line 131, in worker_process_entrypoint
      model = init_model_cb()
    File "/home/ubuntu/taoz/llama-stack/llama_toolchain/inference/meta_reference/model_parallel.py", line 48, in init_model_cb
      llama = Llama.build(config)
    File "/home/ubuntu/taoz/llama-stack/llama_toolchain/inference/meta_reference/generation.py", line 100, in build
      assert len(checkpoints) > 0, f"no checkpoint files found in {ckpt_dir}"
  AssertionError: no checkpoint files found in /home/ubuntu/.llama/checkpoints/Meta-Llama3.1-405B-Instruct/original

Under the original folder,

ubuntu@node007:/mnt/nfs_share/taoz/.llama/checkpoints/Meta-Llama3.1-405B-Instruct/original$ ls
consolidated.00  consolidated.02  consolidated.04  consolidated.06  fp8_scales_0.pt  fp8_scales_2.pt  fp8_scales_4.pt  fp8_scales_6.pt  params.json  tokenizer.model
consolidated.01  consolidated.03  consolidated.05  consolidated.07  fp8_scales_1.pt  fp8_scales_3.pt  fp8_scales_5.pt  fp8_scales_7.pt  README.md

consolidated.xx are folders instead of files, I think that's probably why they were not found.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants
@dawenxi-007 and others