# T5 omits some characters

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ft5-omits-some-characters%2F76592%3E%5C
**Category:** 🤗Transformers\
**Created:** [March 8, 2024, 10:49pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ft5-omits-some-characters%2F76592 "2024-03-08T22:49:13Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![daelba](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/daelba/32/24833_2.png) [@daelba](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdaelba)%5C
**Post date:** [March 8, 2024, 10:49pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ft5-omits-some-characters%2F76592%2F1 "2024-03-08T22:49:14Z")

</div>

I am trying to refine T5 model. My training set is like:

input: “wd:Absolon, Johann, Violinvirtuose, \* 18. 7. 1747 Auscha Bez. Leitmeritz (Úštěk), † 12. 10. 1810 Petersburg. Ausbildung in Prag. Diskantist bei den Kreuzherrn; Geiger in der kaiserlichen Hofmusikkapelle in St. Petersburg bis etwa 1796/97.”

output: “{{WD|label|Absolon, Johann}}, {{WD|P106||Violinvirtuose}}, \* {{WD|P569|1747-07-18|18. 7. 1747}} {{WD|P19|Q330717|Auscha Bez. Leitmeritz (Úštěk)}}, † {{WD|P570|1810-10-12|12. 10. 1810}} {{WD|P20|Q656|Petersburg}}. {{WD|P69|?|qual1=P276:Q1085|Ausbildung in Prag}}. {{WD|P106||Diskantist bei den Kreuzherrn}}; {{WD|P106|Q1259917|qual1=P276:Q656|qual2=P582:1796|Geiger in der kaiserlichen Hofmusikkapelle in St. Petersburg bis etwa 1796/97}}.”

However, after training, I get results like: WD|label|Absolon, Johann, WD|P106||Violinvirtuose, \* WD|P569|1747-07-18 …

When I try to train it with output using “[[xxx]]” like: “[[WD|label|Absolon, Johann]]” I get the right result.

Why is it so? Is there any way to train it with “{{xxx}}”?

---

<div class="post-metadata">

**Author:** ![daelba](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/daelba/32/24833_2.png) [@daelba](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdaelba)%5C
**Post date:** [March 12, 2024, 8:24pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ft5-omits-some-characters%2F76592%2F2 "2024-03-12T20:24:45Z")

</div>

Using mT5 solved this problem.
