skills.rest
MLT-OSS
MLT-OSSOFFICIAL
Explore all skills by this author →
WebsiteGitHub

axolotl

Configure and fine-tune LLMs with Axolotl YAML configs, LoRA, and RLHF methods.

14|5|Updated Apr 9, 2026
Search Tags:#llm#fine-tuning#lora#qlora#dpo#axolotl#yaml config
One-click install
npx skills add https://github.com/MLT-OSS/hermes-agent-go --skill axolotl-mlt-oss
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: axolotl
Source: https://github.com/MLT-OSS/hermes-agent-go/tree/main/skills/mlops/training/axolotl
Command: npx skills add https://github.com/MLT-OSS/hermes-agent-go --skill axolotl-mlt-oss

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Fine-tuning large language models involves complex configuration of training methods, dataset formats, distributed training, and optimization techniques. This Skill provides expert guidance on Axolotl so you can write correct YAML configs and debug training issues without digging through hundreds of documentation pages. ## Core Features & Use Cases - YAML Config Guidance: Covers LoRA/QLoRA, full fine-tuning, FSDP, DeepSpeed, mixed precision (FP16/BF16/FP8), and quantization with torchao. - Dataset Format Support: Explains pre-training, instruction, conversation (chat_template), template-free, and preference dataset formats for DPO, KTO, ORPO, and GRPO. - API & Troubleshooting Reference: Includes 150+ API pages plus an FAQ covering NCCL, DeepSpeed, chat template, and tokenization errors. - Use Case: When setting up QLoRA fine-tuning of a Llama model on a custom chat dataset, use this Skill to generate the correct YAML config with chat_template, roles_to_train, and FSDP settings. ## Quick Start Ask the agent to write an Axolotl YAML config for QLoRA fine-tuning of a Llama model on your instruction dataset.

Frequently Asked Questions about axolotl

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fine-tune an LLM with Axolotl?▼

Fine-tuning with Axolotl works by writing a YAML config that specifies the base model, dataset paths and types, adapter method such as LoRA or QLoRA, and training hyperparameters, then running the axolotl train CLI command. The config-driven approach avoids writing custom training code.

What dataset formats does Axolotl support?▼

Axolotl supports pre-training text datasets, instruction formats like Alpaca, conversation formats using chat_template with OpenAI or ShareGPT-style messages, template-free input_output formats with segment masking, and preference formats for DPO, KTO, ORPO, and GRPO.

Does Axolotl support QLoRA and LoRA fine-tuning?▼

Yes, Axolotl supports LoRA and QLoRA via the adapter config key, with options like lora_target_modules and 4-bit or 8-bit loading. It also includes optimized Triton kernels for LoRA MLP and attention computations to improve speed and memory usage.

Can Axolotl train multimodal or vision language models?▼

Axolotl has beta multimodal support for models like Llama 3.2 Vision, Llama 4, Pixtral, and Llava using an extended chat_template format with image, audio, or video content. Sample packing is not yet supported and some architectures have limitations.

Why does Axolotl training hang or fail with DeepSpeed?▼

Hanging is often caused by GPU communication issues, which NCCL tests can diagnose. Exit code -7 usually means upgrading DeepSpeed helps, and single-GPU runs should remove the deepspeed config section to avoid mpi4py module errors.

How do I fix chat template errors in Axolotl preprocessing?▼

Chat template errors like missing content boundaries usually mean the eos_token does not match the template or message_property_mappings are misconfigured. Verify role and content key mappings and set special_tokens such as pad_token and eos_token explicitly in the YAML.

Related Skills

Explore More Software Engineering Skills→
Dianshu-LiaoDianshu-Liao
2

hugging-face-dataset-creator

Create and manage Hugging Face dataset repositories with validated JSONL training examples.

Community
Advanced
cuba6112cuba6112

unsloth-gguf

Export fine-tuned machine learning models to GGUF format with quantization and LoRA merging.

Community
Intermediate
NVIDIANVIDIA
49

sop-cr-finetuning

Launch and monitor Cosmos-Reason2 fine-tuning on augmented video QA datasets.

Official
Advanced
HuskyDannyHuskyDanny

unsloth

Fine-tune LLMs with LoRA and QLoRA using the Unsloth framework.

Community
Advanced