File size: 305 Bytes
069f69c |
1 2 3 |
* 20240525: Pushed v0.4 LoRA, even more deslopping/dataset fixes. * 20240521: Pushed v0.3 LoRA, with more deslopping, and dataset fixes. Also uses HQQ quantization instead of QLoRA, which got training up to 8k context size. * 20240511: Pushed v0.2 LoRA. Higher context size, and curation/fixes to dataset. |