I can not run the recipe eleuther_eval

[MaxwelsDonc]

I finetune llama3 model and I try to evaluate it by running the recipes eleuther_eval. Here is my command tune run eleuther_eval --config ./eval_config.yaml and the yaml file are as follows:

model:
  _component_: torchtune.models.llama3.llama3_8b

checkpointer:
  _component_: torchtune.utils.FullModelMetaCheckpointer
  checkpoint_dir: Meta-Llama-3-8B
  checkpoint_files: [meta_model_0.pt]
  recipe_checkpoint: null
  output_dir: Meta-Llama-3-8B
  model_type: LLAMA3

# Tokenizer
tokenizer:
  _component_: torchtune.models.llama3.llama3_tokenizer
  path: Meta-Llama-3-8B/original/tokenizer.model

# Environment
device: cuda
dtype: bf16
seed: 217

# EleutherAI specific eval args
tasks: ["gsm8k_cot"]
limit: null
max_seq_length: 4096

# Quantization specific args
quantizer: null

The directory in the yaml file do not have problems. But when I run the command, the output is

2024-05-16:00:05:54,930 INFO     [_utils.py:34] Running EleutherEvalRecipe with resolved config:

checkpointer:
  _component_: torchtune.utils.FullModelMetaCheckpointer
  checkpoint_dir: Meta-Llama-3-8B
  checkpoint_files:
  - meta_model_0.pt
  model_type: LLAMA3
  output_dir: Meta-Llama-3-8B
  recipe_checkpoint: null
device: cuda
dtype: bf16
limit: null
max_seq_length: 4096
model:
  _component_: torchtune.models.llama3.llama3_8b
quantizer: null
seed: 217
tasks:
- gsm8k_cot
tokenizer:
  _component_: torchtune.models.llama3.llama3_tokenizer
  path: Meta-Llama-3-8B/original/tokenizer.model

2024-05-16:00:05:55,066 DEBUG    [seed.py:59] Setting manual seed to local seed 217. Local seed is seed + rank = 217 + 0
2024-05-16:00:05:58,121 INFO     [eleuther_eval.py:168] Model is initialized with precision torch.bfloat16.
2024-05-16:00:05:58,583 INFO     [eleuther_eval.py:152] Tokenizer is initialized from file.
Traceback (most recent call last):
  File "/home/zzh/miniconda3/envs/torchtune/bin/tune", line 8, in <module>
    sys.exit(main())
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/torchtune/_cli/tune.py", line 49, in main
    parser.run(args)
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/torchtune/_cli/tune.py", line 43, in run
    args.func(args)
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/torchtune/_cli/run.py", line 179, in _run_cmd
    self._run_single_device(args)
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/torchtune/_cli/run.py", line 93, in _run_single_device
    runpy.run_path(str(args.recipe), run_name="__main__")
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/runpy.py", line 265, in run_path
    return _run_module_code(code, init_globals, run_name,
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/runpy.py", line 97, in _run_module_code
    _run_code(code, mod_globals, init_globals,
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/recipes/eleuther_eval.py", line 211, in <module>
    sys.exit(recipe_main())
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/torchtune/config/_parse.py", line 50, in wrapper
    sys.exit(recipe_main(conf))
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/recipes/eleuther_eval.py", line 207, in recipe_main
    recipe.evaluate()
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/recipes/eleuther_eval.py", line 175, in evaluate
    model_eval_wrapper = _EvalWrapper(
  File "/home/zzh/miniconda3/envs/torchtune/lib/python3.8/site-packages/recipes/eleuther_eval.py", line 58, in __init__
    super().__init__(device=str(device))
TypeError: __init__() missing 1 required positional argument: 'pretrained'

I have look through the source code in here, But I can't debug the code, so do you know why I always encounter such a problem?

[joecummings]

Hi! Eleuther recently updated their code to require this "pretrained" argument. We updated our code. Can you either install from source or from our nightlies?