text-generation-webui/docs/What Works.md
2024-05-21 13:32:02 -03:00

1.6 KiB

What Works

Loader Loading 1 LoRA Loading 2 or more LoRAs Training LoRAs Multimodal extension Perplexity evaluation
Transformers ** *
llama.cpp use llamacpp_HF
llamacpp_HF
ExLlamav2_HF
ExLlamav2 use ExLlamav2_HF
AutoGPTQ
AutoAWQ ? ? ?
HQQ ? ? ? ?

= not implemented

= implemented

* Training LoRAs with GPTQ models also works with the Transformers loader. Make sure to check "auto-devices" and "disable_exllama" before loading the model.

** Multi-LoRA in PEFT is tricky and the current implementation does not work reliably in all cases.