NameError: name 'init_empty_weights' is not defined

[0xarcher]

Traceback (most recent call last):
File "/Users/archer/.pyenv/versions/3.11.9/lib/python3.11/threading.py", line 1045, in _bootstrap_inner
self.run()
File "/Users/archer/.pyenv/versions/3.11.9/lib/python3.11/threading.py", line 982, in run
self._target(*self._args, **self._kwargs)
File "/Users/archer/coding/AI/DeepGit/app.py", line 177, in run_workflow
result = graph.invoke(initial_state)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/langgraph/pregel/init.py", line 1940, in invoke
for chunk in self.stream(
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/langgraph/pregel/init.py", line 1660, in stream
for _ in runner.tick(
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/langgraph/pregel/runner.py", line 167, in tick
run_with_retry(
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/langgraph/pregel/retry.py", line 40, in run_with_retry
return task.proc.invoke(task.input, config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/langgraph/utils/runnable.py", line 408, in invoke
input = step.invoke(input, config, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/langgraph/utils/runnable.py", line 184, in invoke
ret = context.run(self.func, input, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/tools/dense_retrieval.py", line 13, in hybrid_dense_retrieval
sem_model = SentenceTransformer(agent_config.sem_model_name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/sentence_transformers/SentenceTransformer.py", line 308, in init
modules, self.module_kwargs = self._load_sbert_model(
^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/sentence_transformers/SentenceTransformer.py", line 1739, in _load_sbert_model
module = module_class(model_name_or_path, cache_dir=cache_folder, backend=self.backend, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/sentence_transformers/models/Transformer.py", line 81, in init
self._load_model(model_name_or_path, config, cache_dir, backend, is_peft_model, **model_args)
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/sentence_transformers/models/Transformer.py", line 181, in _load_model
self.auto_model = AutoModel.from_pretrained(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/transformers/models/auto/auto_factory.py", line 571, in from_pretrained
return model_class.from_pretrained(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/transformers/modeling_utils.py", line 279, in _wrapper
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/transformers/modeling_utils.py", line 4333, in from_pretrained
model_init_context = cls.get_init_context(is_quantized, _is_ds_init_called)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/archer/coding/AI/DeepGit/venv/lib/python3.11/site-packages/transformers/modeling_utils.py", line 3736, in get_init_context
init_contexts = [no_init_weights(), init_empty_weights()]
^^^^^^^^^^^^^^^^^^
NameError: name 'init_empty_weights' is not defined

[BigRomanov]

The issue is coming from a huggingface/transformers package: huggingface/transformers#37326. Workaround here: huggingface/transformers#37326 (comment)

[zamalali]

this traceback comes from transformers calling init_empty_weights, which is provided by the accelerate package. it means accelerate is either missing from the environment or its installed version is incompatible with the installed transformers version, so the symbol is never imported. fix: pip install -U accelerate, then restart the app. if the error persists after upgrading, pin a known-good pair such as transformers==4.40.2 with accelerate==0.30.1, or downgrade transformers to a version that matches your accelerate. the upstream issue at huggingface/transformers#37326 confirms the same root cause and workaround. accelerate is currently not listed in requirements.txt, so adding it explicitly is a sensible follow-up to prevent this for new users.