In the short term, we are investing in Turbine to make it feature complete and usable for:
- AOT export (nod-ai#125)
- Serving SHARK Gen-AI Models (nod-ai#119)
- Integration with Quantization Workflows (nod-ai#120)
- General eager execution (nod-ai#105)
- FX Importer Completion (nod-ai#139)
We are also investing in project infrastructure:
- Releasing (nod-ai#121)
- CI and Testing (nod-ai#122)
- Upstreaming (nod-ai#123)
- User Documentation and Samples (nod-ai#124)
- Developer/Debugging Workflows (nod-ai#75)
Roughly, we will aim to align delivery of these features with the next official PyTorch release, at which point we will track PyTorch versions as our versioning scheme.