# Set\_seed and training argument's data\_seed **URL:** \ **Category:** Beginners\ **Created:** [August 6, 2024, 3:00am UTC](https://discuss.huggingface.co/t/set-seed-and-training-arguments-data-seed/101048 "2024-08-06T03:00:50Z")\ **Posts on this page:** 3\ **Page:** 1
**Author:** ![lIlBrother](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/lilbrother/32/30226_2.png) [@lIlBrother](https://discuss.huggingface.co/u/lIlBrother)\ **Post date:** [August 6, 2024, 3:00am UTC](https://discuss.huggingface.co/t/set-seed-and-training-arguments-data-seed/101048/1 "2024-08-06T03:00:50Z")
If I specify the set\_seed in the trainer utility, does it also fix the data\_seed in TrainingArguments? To ensure reproducibility, is using set\_seed alone sufficient, or should I also define data\_seed? ---
**Author:** ![lIlBrother](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/lilbrother/32/30226_2.png) [@lIlBrother](https://discuss.huggingface.co/u/lIlBrother)\ **Post date:** [August 6, 2024, 3:04am UTC](https://discuss.huggingface.co/t/set-seed-and-training-arguments-data-seed/101048/2 "2024-08-06T03:04:20Z")
data\_seed (`int`, _optional_): Random seed to be used with data samplers. If not set, random generators for data sampling will use the same seed as `seed`. This can be used to ensure reproducibility of data sampling, independent of the model seed. ---
**Author:** ![system](https://us1.discourse-cdn.com/hellohellohello/original/2X/d/de4155eb4aa4108ecb32a1389d7cc37ae69f88b7.png) [@system](https://discuss.huggingface.co/u/system)\ **Post date:** [August 6, 2024, 3:04pm UTC](https://discuss.huggingface.co/t/set-seed-and-training-arguments-data-seed/101048/3 "2024-08-06T15:04:43Z")
Closed on August 6, 2024, 3:04pm UTC This topic was automatically closed 12 hours after the last reply. New replies are no longer allowed.