Add .gguf
#1
by
Nekochu
- opened
This is a version of checkpoint 2.0 - AI Dungeon 2 Unleashed > https://github.com/AIDungeon/AIDungeon/files/3935881/model_v5.torrent.zip converted to .gguf to keep a trace of the old model. Q8 have been tested in Oobabooga.
- Conversion: TensorFlow 1.15 Checkpoint (model_v5 is a legacy TF1 model) β Hugging Face Transformers (PyTorch) β llama.cpp (GGUF Q8)
- Fix: custom vocab (6400)+
wteto match modern PyTorch shapes.