v0.12.1
v0.12.1 (2024-05-14)
Performance
- perf: use best available torch backend for embedding
Automatically detect and use the best available backend for PyTorch. This is used to specify the device on which to compute the embeddings. Currently, this searches in order:
- CUDA
- MPS
If no device is available, this process will fall back on the CPU. (59dcc06
)