# Will Trainer loss functions automatically ignore -100?
**URL:**
**Author:**  [@PolarisRisingWar](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FPolarisRisingWar)%5C
**Post date:** [April 10, 2023, 1:01pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fwill-trainer-loss-functions-automatically-ignore-100%2F36134%2F1 "2023-04-10T13:01:43Z")
I’m learning the tutorial [Token classification](https://huggingface.co/docs/transformers/tasks/token_classification).
And I have a question:
In preprocessing part, it converted the label of special tokens and the tokens that are not the first token in the word to -100.
It said that this will show its usefulness in loss function, but the code hasn’t use something like `ignore_index` in loss function. The -100 seems only work in metric evaluation.
So I wonder if loss function in Trainer will automatically ignore -100?
Or it works in num\_labels, id2label and label2id parameter? Does it mean that I set these parameters, other labels will be ignored?
---
**Author:**  [@mdrpanwar](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fmdrpanwar)%5C
**Post date:** [June 29, 2023, 10:54am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fwill-trainer-loss-functions-automatically-ignore-100%2F36134%2F2 "2023-06-29T10:54:01Z")
I have the same question for `GPT2LMHeadModel`.
Edit: `-100` is the default `ignore_index` in [PyTorch’s CrossEntropyLoss](https://pytorch.org/docs/stable/generated/torch.nn.CrossEntropyLoss.html). So, any token with a label of `-100` will be ignored in loss computation.
---
**Author:**  [@system](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fsystem)%5C
**Post date:** [May 25, 2024, 6:17am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fwill-trainer-loss-functions-automatically-ignore-100%2F36134%2F3 "2024-05-25T06:17:14Z")
This topic was automatically closed 12 hours after the last reply. New replies are no longer allowed.