Automate end-to-end fine-tuning of the Alpamayo-1 VLM on the PAI dataset.
npx skills add https://github.com/NVlabs/alpamayo-recipes --skill alpamayo1-sft
Please help me install this Agent Skill. Skill: alpamayo1-sft Source: https://github.com/NVlabs/alpamayo-recipes/tree/main/recipes/alpamayo1_sft Command: npx skills add https://github.com/NVlabs/alpamayo-recipes --skill alpamayo1-sft
This Skill automates the end-to-end supervised fine-tuning process for the Alpamayo-1 Vision-Learning Model (VLM) on the Physical AI Autonomous Vehicles (PAI) dataset. It removes the need for manual fine-tuning steps and speeds up the process.
uv virtual environment and the required libraries.Use the alpamayo1-sft skill to fine-tune the Alpamayo-1 VLM on the PAI dataset using the default configuration.
High-intent search queries and answers about installing and using this skill.
You can fine-tune the Alpamayo-1 VLM on the PAI dataset by running an automated end-to-end pipeline that handles data preparation, Stage 1 and Stage 2 training, and model evaluation.
Yes, you need DeepSpeed and Hugging Face transformers to train the Alpamayo-1 VLM. The supervised fine-tuning process also requires Hydra for configuration and flash-attn for model training.
The best way to automate VLM fine-tuning for autonomous vehicle datasets is using a pipeline that manages environment setup, data preparation, and multi-stage training without manual intervention.
Yes, you can select a subset of the PAI dataset for supervised VLM fine-tuning. The data preparation commands include options for subsetting data and enabling CoC reasoning labels.
Hydra configuration works with DeepSpeed for vision-learning model training by orchestrating the multi-stage supervised fine-tuning parameters and environment variables automatically.
VLM fine-tuning requires a specific virtual environment setup to manage complex dependencies like flash-attn, DeepSpeed, and torchmetrics, ensuring the training pipeline executes without version conflicts.
NVlabsAutomate Alpamayo-1.5 VLM fine-tuning on PAI or LingoQA datasets.
codeApe-7Guide dataset validation, hyperparameter optimization, and deployment of fine-tuned large language models.
x-TheFoxLog ML experiments, sweeps, and artifacts with WandB.
NagyViktFine-tune computer vision models on Hugging Face Jobs infrastructure.