Axolotl fine-tuning

Published, not hardware-testedociTraining / fine-tune

Axolotl 0.19 for LoRA, QLoRA and full fine-tuning from YAML configs.

Needs 22 GiB VRAM · CUDA on NVIDIA

No GPU big enough is free right now.

New machines appear as hosts connect them.

Be the first host See all machines

No speed benchmarks yet, so we match on price.

Template details

Rating: No ratings yet
Runs: 0
Min VRAM: 24 GB
Backend: CUDA · NVIDIA GPUs
Version: axolotl-workspace-v1.0.0
Image digest: 923e68f198e0…
Health check: SSH probe on the leased endpoint

SSH as `tenant` with your key; tools are on PATH. The upstream image compiles its kernels for sm_90 and newer (TORCH_CUDA_ARCH_LIST 9.0 10.0 10.3 12.0+PTX), so consumer Ampere/Ada cards are UNVERIFIED. Datasets and checkpoints live in /work. Thin AmpleRun wrapper (templates/images/workspace, variant ws-axolotl) over axolotlai/axolotl:0.19.0@sha256:9de7c7a5b8830480a7d2eb3b6d49759586615f5f8eb1126d5df29f8bd9fa324b (digest verified 2026-09-26); NOT BUILT, so image_digest is a placeholder that can be neither published nor rented. Licence: Apache-2.0 (https://github.com/axolotl-ai-cloud/axolotl/blob/main/LICENSE). Needs an NVIDIA driver that supports CUDA 13.0 or newer. Status: not hardware-tested.

We match GPUs on their measured runtime and dedicated VRAM. A hardware report doesn't qualify another vendor's backend.

#fine-tuning #lora #qlora

Reported on real GPUs

People report these themselves. We haven't checked them.

No reports yet.

Reviews

No reviews yet.