examples/flax/language-modeling/run_t5_mlm_flax.py · 71356034232cfce20d8f24df1213944665e7fe08 · chenpangpang / transformers · GitLab

Find file Blame History Permalink

Speedup training by using numpy instead of jnp for batch shuffling (#15963) · 91fb62d0
Yeb Havinga authored Mar 08, 2022
```
Speedup training by using numpy instead of jnp for batch shuffling
Co-authored-by: Yeb Havinga <y.t.havinga@mgrid.net>
```
91fb62d0

run_t5_mlm_flax.py 39.7 KB