Axolotl

If you're fine-tuning LLMs with Axolotl, this skill gives you direct guidance on YAML configs, LoRA/QLoRA setups, and alignment techniques like DPO and ORPO. It covers practical stuff like FSDP configuration, context parallelism settings, and NCCL performance testing. The quick reference includes actual config snippets and the references folder has detailed docs on dataset formats and API usage. Honestly, Axolotl's YAML-based approach is refreshingly straightforward compared to writing training loops from scratch, and having the multimodal support patterns documented here saves you from piecing together examples from GitHub issues.

skills.sh installs: 1K
GitHub stars: 14.4K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add orchestra-research/ai-research-skills --skill axolotl --agent claude-code

Installs into .claude/skills of the current project.

Files

Axolotl Skill

Comprehensive assistance with axolotl development, generated from official documentation.

When to Use This Skill

This skill should be triggered when:

Working with axolotl
Asking about axolotl features or APIs
Implementing axolotl solutions
Debugging axolotl code
Learning axolotl best practices

Quick Reference

Common Patterns

Pattern 1: To validate that acceptable data transfer speeds exist for your training job, running NCCL Tests can help pinpoint bottlenecks, for example:

./build/all_reduce_perf -b 8 -e 128M -f 2 -g 3

Pattern 2: Configure your model to use FSDP in the Axolotl yaml. For example:

fsdp_version: 2
fsdp_config:
  offload_params: true
  state_dict_type: FULL_STATE_DICT
  auto_wrap_policy: TRANSFORMER_BASED_WRAP
  transformer_layer_cls_to_wrap: LlamaDecoderLayer
  reshard_after_forward: true

Pattern 3: The context_parallel_size should be a divisor of the total number of GPUs. For example: