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?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Fixing the random seed in the Trainer does not produce the same results across runs | 5 | 17695 | March 27, 2025 | |
| Random seed for weight initialization and data order | 0 | 1237 | February 21, 2022 | |
| Set_seed is not working | 2 | 635 | September 8, 2024 | |
| TrainingArguments seed - possible values | 0 | 294 | March 29, 2021 | |
| Is the trainer's seed reset at every model_init? | 4 | 1255 | March 28, 2022 |