Releases: codekansas/ml-pretrained
Releases · codekansas/ml-pretrained
v0.0.18
v0.0.17
- Demucs model
- Small fixes
v0.0.13
llama checkpointing (#41) * bump version * llama checkpointing
v0.0.12
lora improvements (#39)
v0.0.11
freeze options (#38) * freeze options * fix lint * more lint * more lint
v0.0.10
- HuBERT model with K-means
- Remove some dependencies
v0.0.9
- Better HuBERT API
- Triton kernel for RWKV (still not working for training though)
v0.0.8
rwkv return logits (#17)
v0.0.5
stupid fix (#11)
v0.0.4
add py.typed (#10) * add py.typed * bump version