Guide LLaMA-Factory fine-tuning and deployment workflows for large language models.
npx skills add https://github.com/ricable/mcai --skill llama-factory-ricable
Please help me install this Agent Skill. Skill: llama-factory Source: https://github.com/ricable/mcai/tree/main/.agents/skills/llama-factory Command: npx skills add https://github.com/ricable/mcai --skill llama-factory-ricable
This Skill turns LLaMA-Factory documentation into practical guidance for fine-tuning, evaluating, and serving large language models without forcing you to piece together setup steps on your own.
Ask for a step-by-step LLaMA-Factory plan for my model, dataset, and hardware, including the right training stage, adapter setup, and launch command.
High-intent search queries and answers about installing and using this skill.
Fine-tune large language models by configuring supervised fine-tuning, pretraining, or preference optimization stages with LLaMA-Factory. Specify your dataset, model, and hardware to generate the correct adapter setup and launch command for end-to-end tuning workflows.
Yes, LLaMA-Factory supports multimodal training on both CUDA and Ascend NPU hardware. You can configure multimodal runs for models like Qwen or LLaVA by specifying your hardware environment and generating the appropriate distributed execution commands.
LoRA and QLoRA are adapter methods for efficient large language model tuning. LoRA applies low-rank adaptation to reduce trainable parameters, while QLoRA combines quantization with LoRA to further minimize memory usage during distributed training and acceleration workflows.
Prepare Alpaca or ShareGPT datasets for reward modeling by formatting your input data to match LLaMA-Factory templates. The workflow guides dataset configuration and adapter setup to ensure your preference optimization data aligns with the training stage requirements.
Yes, LLaMA-Factory supports Direct Preference Optimization (DPO) and KTO workflows. It provides configuration guidance for reward modeling and preference optimization stages, allowing you to align large language models with human preferences during the fine-tuning process.
Quantization reduces memory usage for large language model inference but may impact model accuracy. LLaMA-Factory provides configuration guidance for quantization and acceleration options, helping you balance efficiency trade-offs when deploying fine-tuned models in production environments.
FeRhodiumAnalyze PyTorch model architectures for Ascend NPU migration requirements.
Simon-Copilot-StudioFine-tune large language models with LoRA and QLoRA adapters.
DevelopersGlobalImplement structured logging, distributed tracing, and alerting with OpenTelemetry.
cxnaiveGenerate and manage YAML configurations for Axolotl fine-tuning of LLMs.
