An error when running fine-tuning-embeddinggemma-with-sentence-transformers.ipynb on kaggle

[linsong8208]

Description of the bug:

Everything goes on well until in the "training" part, instead of training starts, the error is like:
TypeError Traceback (most recent call last)
/tmp/ipykernel_55/2596833136.py in <cell line: 0>()
39 callbacks=[MyCallback(evaluate)]
40 )
---> 41 trainer.train()

/usr/local/lib/python3.12/dist-packages/transformers/trainer.py in train(self, resume_from_checkpoint, trial, ignore_keys_for_eval, **kwargs)
2326 hf_hub_utils.enable_progress_bars()
2327 else:
-> 2328 return inner_training_loop(
2329 args=args,
2330 resume_from_checkpoint=resume_from_checkpoint,

/usr/local/lib/python3.12/dist-packages/transformers/trainer.py in _inner_training_loop(self, batch_size, args, resume_from_checkpoint, trial, ignore_keys_for_eval)
2670 )
2671 with context():
-> 2672 tr_loss_step = self.training_step(model, inputs, num_items_in_batch)
2673
2674 if (

/usr/local/lib/python3.12/dist-packages/transformers/trainer.py in training_step(self, model, inputs, num_items_in_batch)
4018
4019 with self.compute_loss_context_manager():
-> 4020 loss = self.compute_loss(model, inputs, num_items_in_batch=num_items_in_batch)
4021
4022 del inputs

/usr/local/lib/python3.12/dist-packages/sentence_transformers/base/trainer.py in compute_loss(self, model, inputs, return_outputs, num_items_in_batch)
471 ):
472 loss_fn = self.override_model_in_loss(loss_fn, model)
--> 473 loss = loss_fn(features, labels)
474 if isinstance(loss, dict):
475 self.track_loss_components(loss)

/usr/local/lib/python3.12/dist-packages/torch/nn/modules/module.py in _wrapped_call_impl(self, *args, **kwargs)
1774 return self._compiled_call_impl(*args, **kwargs) # type: ignore[misc]
1775 else:
-> 1776 return self._call_impl(*args, **kwargs)
1777
1778 # torchrec tests the code consistency with the following code

/usr/local/lib/python3.12/dist-packages/torch/nn/modules/module.py in _call_impl(self, *args, **kwargs)
1785 or _global_backward_pre_hooks or _global_backward_hooks
1786 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1787 return forward_call(*args, **kwargs)
1788
1789 result = None

/usr/local/lib/python3.12/dist-packages/sentence_transformers/sentence_transformer/losses/multiple_negatives_ranking.py in forward(self, sentence_features, labels)
230 def forward(self, sentence_features: Iterable[dict[str, Tensor]], labels: Tensor) -> Tensor:
231 # Compute the embeddings and distribute them to anchor and candidates (positive and optionally negatives)
--> 232 embeddings = [self.model(sentence_feature)["sentence_embedding"] for sentence_feature in sentence_features]
233 return self.compute_loss_from_embeddings(embeddings, labels)
234

/usr/local/lib/python3.12/dist-packages/torch/nn/modules/module.py in _wrapped_call_impl(self, *args, **kwargs)
1774 return self._compiled_call_impl(*args, **kwargs) # type: ignore[misc]
1775 else:
-> 1776 return self._call_impl(*args, **kwargs)
1777
1778 # torchrec tests the code consistency with the following code

/usr/local/lib/python3.12/dist-packages/torch/nn/modules/module.py in _call_impl(self, *args, **kwargs)
1785 or _global_backward_pre_hooks or _global_backward_hooks
1786 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1787 return forward_call(*args, **kwargs)
1788
1789 result = None

/usr/local/lib/python3.12/dist-packages/torch/nn/parallel/data_parallel.py in forward(self, *inputs, **kwargs)
196 replicas = self.replicate(self.module, self.device_ids[: len(inputs)])
197 outputs = self.parallel_apply(replicas, inputs, module_kwargs)
--> 198 return self.gather(outputs, self.output_device)
199
200 def replicate(self, module: T, device_ids: Sequence[int | torch.device]) -> list[T]:

/usr/local/lib/python3.12/dist-packages/torch/nn/parallel/data_parallel.py in gather(self, outputs, output_device)
217
218 def gather(self, outputs: Any, output_device: int | torch.device) -> Any:
--> 219 return gather(outputs, output_device, dim=self.dim)
220
221

/usr/local/lib/python3.12/dist-packages/torch/nn/parallel/scatter_gather.py in gather(outputs, target_device, dim)
149 # Setting the function to None clears the refcycle.
150 try:
--> 151 res = gather_map(outputs)
152 finally:
153 gather_map = None # type: ignore[assignment]

/usr/local/lib/python3.12/dist-packages/torch/nn/parallel/scatter_gather.py in gather_map(outputs)
139 raise ValueError("All dicts must have the same number of keys")
140 # pyrefly: ignore [not-callable]
--> 141 return type(out)((k, gather_map([d[k] for d in outputs])) for k in out)
142 if _is_namedtuple(out):
143 # pyrefly: ignore [no-matching-overload]

/usr/local/lib/python3.12/dist-packages/torch/nn/parallel/scatter_gather.py in (.0)
139 raise ValueError("All dicts must have the same number of keys")
140 # pyrefly: ignore [not-callable]
--> 141 return type(out)((k, gather_map([d[k] for d in outputs])) for k in out)
142 if _is_namedtuple(out):
143 # pyrefly: ignore [no-matching-overload]

/usr/local/lib/python3.12/dist-packages/torch/nn/parallel/scatter_gather.py in gather_map(outputs)
144 return type(out)._make(map(gather_map, zip(*outputs, strict=True)))
145 # pyrefly: ignore [no-matching-overload]
--> 146 return type(out)(map(gather_map, zip(*outputs, strict=True)))
147
148 # Recursive function calls like this create reference cycles.

TypeError: 'int' object is not iterable

Actual vs expected behavior:

No response

Any other information you'd like to share?

No response