How to free up tensor from GPU while using en_core_web_trf for inference
Open
feat / transformer
gpu
perf / memory
- Dominant language
- Python
- Stars
- 33.9k
- Forks
- 4.7k
- Avg merge
- 3m
- Merged PRs (30d)
- 1
Description
Getting OOM while inferencing documents using the spacy Transformer model.
The GPU memory is not getting free up when doc object get created.
Contributor guide
Assessment
This issue has not been assessed yet.