mistral.rs
mistral.rs copied to clipboard
Tracking issue for AnyMoE
This is a tracking issue for the development of AnyMoE, which will be broken up into several PRs.
- [x] Core functionality, plain models, all APIs: #476
- [x] Support for vision models: #515
- [ ] Support saving and loading gating layer: #519
- [ ] Generate graph of loss
- [ ] Read dataset from parquet instead of CSV