Loss is “nan” when fine-tuning HuggingFace NLI model (both RoBERTa/BART)

Question

I'm using HuggingFace's Transformer's library and I’m trying to fine-tune a pre-trained NLI model (ynie/roberta-large-snli_mnli_fever_anli_R1_R2_R3-nli) on a dataset of around 276.000 hypothesis-premise pairs. I’m following the instructions from the docs here and here. I have the impression that the fine-tuning works (it does the training and saves the checkpoints), but trainer.train() and trainer.evaluate() return "nan" for the loss.

What I've tried:

• I tried using both ynie/roberta-large-snli_mnli_fever_anli_R1_R2_R3-nli and facebook/bart-large-mnli to make sure that it's not linked to specific model, but I get the issue for both models
• I tried following the advice in this related github issue, but adding num_labels=3 to the config file does not solve the issue. (I think my issue is different because the models are already fine-tuned on NLI in my case)
• I tried many different ways of changing my input data because I suspected that there could be an issue with my input data, but I also couldn't solve it that way.
• The probable source of the issue: I inspected the prediction output from the model during training and the weird thing is the prediction value always seems to be "0" (entailment) in 100% of cases (see printed output in the code below). This is clearly an error. I think the source for this is that the logits the model seems to return during training are torch.tensor([[np.nan, np.nan, np.nan]]) and when you apply .argmax(-1) to this, you get torch.tensor(0). The big mystery for me is why the logits would become "nan", because the model does not do that when I use the same input data only outside of the trainer. => Does anyone know where this issues comes from? See my code below.

Thanks a lot in advance for any suggestion!

Here is my code:

### load model & tokenize
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

max_length = 256
hg_model_hub_name = "ynie/roberta-large-snli_mnli_fever_anli_R1_R2_R3-nli"
# also tried: hg_model_hub_name = "facebook/bart-large-mnli"
tokenizer = AutoTokenizer.from_pretrained(hg_model_hub_name)
model = AutoModelForSequenceClassification.from_pretrained(hg_model_hub_name)
model.config

device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Device: {device}")
if device == "cuda":
  model = model.half()
model.to(device)
model.train();

#... some data preprocessing

encodings_train = tokenizer(premise_train, hypothesis_train, return_tensors="pt", max_length=max_length,
                            return_token_type_ids=True, truncation=False, padding=True)
encodings_val = tokenizer(premise_val, hypothesis_val, return_tensors="pt", max_length=max_length,
                          return_token_type_ids=True, truncation=False, padding=True)
encodings_test = tokenizer(premise_test, hypothesis_test, return_tensors="pt", max_length=max_length,
                           return_token_type_ids=True, truncation=False, padding=True)


### create pytorch dataset object
class XDataset(torch.utils.data.Dataset):
    def __init__(self, encodings, labels):
        self.encodings = encodings
        self.labels = labels
    def __getitem__(self, idx):
        item = {key: torch.as_tensor(val[idx]) for key, val in self.encodings.items()}
        #item = {key: torch.as_tensor(val[idx]).to(device) for key, val in self.encodings.items()}
        item['labels'] = torch.as_tensor(self.labels[idx])
        #item['labels'] = self.labels[idx]
        return item
    def __len__(self):
        return len(self.labels)

dataset_train = XDataset(encodings_train, label_train)
dataset_val = XDataset(encodings_val, label_val)
dataset_test = XDataset(encodings_test, label_test)

# compute metrics with trainer
from sklearn.metrics import accuracy_score, precision_recall_fscore_support
def compute_metrics(pred):
    labels = pred.label_ids
    print(labels)
    preds = pred.predictions.argmax(-1)
    print(preds)
    precision, recall, f1, _ = precision_recall_fscore_support(labels, preds, average='binary', pos_label=0)
    acc = accuracy_score(labels, preds)
    return {
        'accuracy': acc,
        'f1': f1,
        'precision': precision,
        'recall': recall
    }


## training
from transformers import Trainer, TrainingArguments

# https://huggingface.co/transformers/main_classes/trainer.html#transformers.TrainingArguments
training_args = TrainingArguments(
    output_dir='./results',          # output directory
    num_train_epochs=1,              # total number of training epochs
    per_device_train_batch_size=8,  # batch size per device during training
    per_device_eval_batch_size=8,   # batch size for evaluation
    warmup_steps=500,                # number of warmup steps for learning rate scheduler
    weight_decay=0.01,               # strength of weight decay
    logging_dir='./logs',            # directory for storing logs
    logging_steps=100,
)

trainer = Trainer(
    model=model,                         # the instantiated 🤗 Transformers model to be trained
    args=training_args,                  # training arguments, defined above
    train_dataset=dataset_train,         # training dataset
    eval_dataset=dataset_val             # evaluation dataset
)

trainer.train()
# output: TrainOutput(global_step=181, training_loss=nan)
trainer.evaluate()
# output: 
[2 2 2 0 0 2 2 2 0 2 0 0 2 2 2 2 0 2 0 2 2 2 2 0 2 0 2 0 0 2 0 0 2 0 0 0 2
 0 2 0 0 0 0 0 2 0 0 2 2 2 0 2 2 2 2 2 0 0 0 0 2 0 0 0 2 2 0 0 0 2 0 0 0 2
 2 0 2 0 0 2 2 2 0 2 2 0 0 0 0 0 0 0 2 0 0 0 0 2 0 2 2 0 2 0 0 2 2 2 2 2 2
 2 0 0 0 0 2 0 0 2 0 0 0 0 2 2 2 0 0 0 0 0 2 0 0 2 0 2 0 2 0 2 0 0 2 2 0 0
 2 2 2 2 2 2 0 0 2 2 2 2 0 2 0 0 2 2 2 0 0 2 0 2 0 2 0 0 0 0 0 0 2 0 0 2 2
 0 2 2 2 0 2 2 0 2 2 2 2 2 2 0 0 2 0 0 2 2 0 0 0 2 0 2 2 2 0 0 0 0 0 0 0 0
 2 0 2 2 2 0 2 0 0 2 0 2 2 0 0 0 0 2 2 2 0 0 0 2 2 2 2 0 2 0 2 2 2]
[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]

{'epoch': 1.0,
 'eval_accuracy': 0.5137254901960784,
 'eval_f1': 0.6787564766839378,
 'eval_loss': nan,
 'eval_precision': 0.5137254901960784,
 'eval_recall': 1.0}

Edit: I've also opened a github issue with a but more detailed description of the issue here: https://github.com/huggingface/transformers/issues/9160

Answer 1

I received a good answer from the HuggingFace team on github. The issue was the model.half(), which has the advantage of increasing speed and reducing memory usage, but it also changes the model in a way that it produces the error. removing the model.half() solved the issue for me. For details, see https://github.com/huggingface/transformers/issues/9160