I am a beginner trying to train Minilm using Hugging Face
I have followed this tutorial: Simple Training with the 🤗 Transformers Trainer - YouTube
My Dataset:
DatasetDict({
train: Dataset({
features: [‘tec_name’, ‘label’, ‘sentence’],
num_rows: 10356
})
validation: Dataset({
features: [‘tec_name’, ‘label’, ‘sentence’],
num_rows: 1295
})
test: Dataset({
features: [‘tec_name’, ‘label’, ‘sentence’],
num_rows: 1294
})
})
Codes:
from transformers import TrainingArguments
batch_size = 64
logging_steps = len(ds[“train”])
output_dir1=“minilm-finetuned_1”
training_args = TrainingArguments(output_dir=output_dir1,
num_train_epochs=5,
learning_rate=2e-5,
per_device_train_batch_size=batch_size,
per_device_eval_batch_size=batch_size,
weight_decay=0.01,
evaluation_strategy=“epoch”,
push_to_hub=True)
trainer = WeightedLossTrainer(model=model,
args=training_args,
compute_metrics=compute_metrics,
train_dataset=ds[“train”],
eval_dataset=ds[“validation”],
tokenizer=tokenizer)
trainer.train()
However an error was returned when I trained the model which showed the following:
/usr/local/lib/python3.10/dist-packages/transformers/optimization.py:407: FutureWarning: This implementation of AdamW is deprecated and will be removed in a future version. Use the PyTorch implementation torch.optim.AdamW instead, or set no_deprecation_warning=True to disable this warning
warnings.warn(
[501/810 00:58 < 00:36, 8.47 it/s, Epoch 3.09/5]
Step Training Loss Validation Loss
[21/21 00:00]
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ in <cell line: 1>:1 │
│ │
│ /usr/local/lib/python3.10/dist-packages/transformers/trainer.py:1664 in train │
│ │
│ 1661 │ │ inner_training_loop = find_executable_batch_size( │
│ 1662 │ │ │ self._inner_training_loop, self._train_batch_size, args.auto_find_batch_size │
│ 1663 │ │ ) │
│ ❱ 1664 │ │ return inner_training_loop( │
│ 1665 │ │ │ args=args, │
│ 1666 │ │ │ resume_from_checkpoint=resume_from_checkpoint, │
│ 1667 │ │ │ trial=trial, │
│ │
│ /usr/local/lib/python3.10/dist-packages/transformers/trainer.py:2019 in _inner_training_loop │
│ │
│ 2016 │ │ │ │ │ self.state.epoch = epoch + (step + 1 + steps_skipped) / steps_in_epo │
│ 2017 │ │ │ │ │ self.control = self.callback_handler.on_step_end(args, self.state, s │
│ 2018 │ │ │ │ │ │
│ ❱ 2019 │ │ │ │ │ self._maybe_log_save_evaluate(tr_loss, model, trial, epoch, ignore_k │
│ 2020 │ │ │ │ else: │
│ 2021 │ │ │ │ │ self.control = self.callback_handler.on_substep_end(args, self.state │
│ 2022 │
│ │
│ /usr/local/lib/python3.10/dist-packages/transformers/trainer.py:2300 in _maybe_log_save_evaluate │
│ │
│ 2297 │ │ │ │ │ ) │
│ 2298 │ │ │ │ │ metrics.update(dataset_metrics) │
│ 2299 │ │ │ else: │
│ ❱ 2300 │ │ │ │ metrics = self.evaluate(ignore_keys=ignore_keys_for_eval) │
│ 2301 │ │ │ self._report_to_hp_search(trial, self.state.global_step, metrics) │
│ 2302 │ │ │ │
│ 2303 │ │ │ # Run delayed LR scheduler now that metrics are populated │
│ │
│ /usr/local/lib/python3.10/dist-packages/transformers/trainer.py:3029 in evaluate │
│ │
│ 3026 │ │ start_time = time.time() │
│ 3027 │ │ │
│ 3028 │ │ eval_loop = self.prediction_loop if self.args.use_legacy_prediction_loop else se │
│ ❱ 3029 │ │ output = eval_loop( │
│ 3030 │ │ │ eval_dataloader, │
│ 3031 │ │ │ description=“Evaluation”, │
│ 3032 │ │ │ # No point gathering the predictions if there are no metrics, otherwise we d │
│ │
│ /usr/local/lib/python3.10/dist-packages/transformers/trainer.py:3318 in evaluation_loop │
│ │
│ 3315 │ │ │ │ │ EvalPrediction(predictions=all_preds, label_ids=all_labels, inputs=a │
│ 3316 │ │ │ │ ) │
│ 3317 │ │ │ else: │
│ ❱ 3318 │ │ │ │ metrics = self.compute_metrics(EvalPrediction(predictions=all_preds, lab │
│ 3319 │ │ else: │
│ 3320 │ │ │ metrics = {} │
│ 3321 │
│ in compute_metrics:5 │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
AttributeError: ‘EvalPrediction’ object has no attribute ‘prediction’
Any assistance is welcomed. Please Help 