Axolotl AI Cloud released Axolotl v0.20.0, comprising 67 commits since v0.19.01. The open-source fine-tuning framework now supports direct GGUF export for llama.cpp, Ollama, LM Studio, and llamafile; Ringmaster context parallelism with experimental support for recurrent layers; native NVFP4 LoRA with merge-aware training under FSDP2, DeepSpeed ZeRO-1/2/3, or tensor parallelism; and expert parallelism without DeepEP. The release removes FSDP1 and raises minimums to Python 3.12 and PyTorch 2.13.0.
Axolotl v0.20.0 Adds GGUF Export, Context Parallelism, NVFP4 LoRA
Axolotl v0.20.0 introduces GGUF export, Ringmaster context parallelism, native NVFP4 LoRA, and expert parallelism while removing FSDP1.
Was this useful?
The day’s vector, in your inbox.
Vector Wire Daily: one dispatch each publishing day.