Releases: freddyheppell/executor-text-transformers-torch-encoder
Releases · freddyheppell/executor-text-transformers-torch-encoder
v0.2: Bump to Jina 3.16.0
- Supports Jina 3.16.0
v0.1: Initial Release
This executor takes Jina's TransformerTorchEncoder, and updates it to:
- support CUDA 11.3
- provide arm64 CPU-only images so the container can run without very slow emulation on M1 macs.
Using from Jina Hub
Flow().add(uses='jinahub+docker://TransformerTorchEncoderCU113/v0.1')
# or
Flow().add(uses='jinahub+docker://TransformerTorchEncoderCU113/v0.1-gpu')
Using from GHCR
# on amd64 or arm64 systems:
Flow().add(uses='docker://ghcr.io/freddyheppell/transformer-torch-encoder:v0.1')
# on amd64 systems:
Flow().add(uses='docker://ghcr.io/freddyheppell/transformer-torch-encoder:v0.1-gpu')
v0.1 Alpha 2
Full Changelog: v0.1-alpha.1...v0.1-alpha.2
v0.1 Alpha 1
Experimental version of CUDA 11.3 compatible executor