快速AI定制

Prerequisites

• NVIDIA GPU using the 7.x or 8.x Compute Capability
• NVIDIA CUDA Toolkit 11.8+
• Python 3.12.x
• PyTorch 2.8.0+ with corresponding forward compatible prebuilt CUDA binaries

For Fine-tuning/Post-Training: Install specific dependencies and initialize rapidfireai

rapidfireai init --train rapidfireai start

For RAG/Context Engineering Evals: Install specific dependencies and initialize rapidfireai

rapidfireai init rapidfireai start

Development prerequisites

TODO: This section needs updating

• Python 3.12.x
• Git
• Ubuntu/Debian system (for apt package manager)

```bash

install dependencies

sudo apt update -y

install basic dependencies

sudo apt install -y python3.12-venv python3 -m venv .venv source .venv/bin/activate pip3 install ipykernel pip3 install jupyter pip3 install "huggingface-hub[cli]" export PATH="$HOME/.local/bin:$PATH" hf auth login --token <your_token>

Getting Started

Install and Get Started

```bash

Forward this port if you installed rapidfireai on a remote machine

ssh -L 8853:localhost:8853 username@remote-machine

Forward these ports if you installed rapidfireai on a remote machine

ssh -L 8850:localhost:8850 -L 8851:localhost:8851 -L 8853:localhost:8853 -L 8852:localhost:8852 username@remote-machine

install the repository as a python package

pip3 install -r requirements.txt

install node

curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash - && sudo apt-get install -y nodejs

Install correct version of vllm and flash-attn

uv pip install vllm=0.10.1.1 --torch-backend=cu126 or cu118

uv pip install flash-attn==1.0.9 --no-build-isoloation or 2.8.3

check installations

node -v # 22.x

Open an example notebook from ./tutorial_notebooks/[fine-tuning | post-training] and start experiment

Open an example notebook from ./tutorial_notebooks/rag-contexteng/ and start experiment

```

still inside venv, run the start script to begin all 3 servers

chmod +x ./rapidfireai/start_dev.sh ./rapidfireai/start_dev.sh start

make sure the notebook is running in the .venv virtual environment

head to settings in Cursor/VSCode and search for venv and add the path - $HOME/rapidfireai/.venv

RapidFireAI Environment Variables

RapidFire AI has sane defaults for most installations, if customization is needed the following operating system variables can be used to overwrite the defaults.

• RF_HOME - Base RapidFire AI home directory (default: ${HOME}/rapidfireai on Non-Google Colab and /content/rapidfireai on Google Colab)
• RF_LOG_PATH - Base directory to store log files (default: ${RF_HOME}/logs)
• RF_EXPERIMENT_PATH - Base directory to store experiment work files (default: ${RF_HOME}/rapidfire_experiments)
• RF_TENSORBOARD_LOG_DIR - Base directory for TensorBoard logs (default: ${RF_EXPERIMENT_PATH}/tensorboard_logs))
• RF_LOG_FILENAME - Default log file name (default: rapidfire.log)
• RF_TRAINING_LOG_FILENAME - Default training log file name (default: training.log)
• RF_DB_PATH - Base directory for database files (default: ${RF_HOME}/db)
• RF_MLFLOW_ENABLED - Enable MLflow tracking backend
• RF_TENSORBOARD_ENABLED - Enable TensorBoard tracking backend
• RF_TRACKIO_ENABLED - Enable Trackio tracking backend
• RF_COLAB_MODE - Whether running on colab (default: false on Non-Google Colab and true on Google Colab)
• RF_TUTORIAL_PATH - Location that rapidfireai init copies tutorial_notebooks to (default: ./tutorial_notebooks)
• RF_TEST_PATH - Location that rapidfireai --test-notebooks copies test notebooks to (default: ./tutorial_notebooks/tests)
• RF_JUPYTER_HOST - Host that rapidfireai jupyter creates a Jupyter listener for (default: 0.0.0.0)
• RF_JUPYTER_PORT - Port that rapidfireai jupyter creates a Jupyter listener for (default: 8850)
• RF_API_HOST - Host that rapidfireai start or Experiment creates an API listener for (default: 0.0.0.0)
• RF_API_PORT - Port that rapidfireai start or Experiment creates an API listener for (default: 8851)
• RF_MLFLOW_HOST - Host that rapidfireai start creates a MLflow listener for (default: 0.0.0.0)
• RF_MLFLOW_PORT - Port that rapidfireai start creates a MLflow listener for (default: 8852)
• RF_FRONTEND_HOST - Host that rapidfireai start creates a Frontend listener for (default: 0.0.0.0)
• RF_FRONTEND_PORT - Port that rapidfireai start creates a Frontend listener for (default: 8853)
• RF_RAY_HOST - Host that Experiment creates a Ray dashboard listener for (default: 0.0.0.0)
• RF_RAY_PORT - Port that Experiment creates a Ray dashboard listener for (default: 8855)
• RF_TIMEOUT_TIME - Time in seconds that services wait to start (default: 30)
• RF_PID_FILE - File to store process ids of started services (default: ${RF_HOME}/rapidfire_pids.txt)
• RF_PYTHON_EXECUTABLE - Python executable (default: python3 falls back to python if not found)
• RF_PIP_EXECUTABLE - pip executable (default: pip3 falls back to pip if not found)
• RF_CONVERGE_MODE - Whether to use Rapidfire AI Converge frontend and backend if available (default: all)
• RF_NO_FRONTEND - Option to disable starting the frontend

RapidFire AI

Rapid experimentation for easier, faster, and more impactful AI customization. Built for agentic RAG, context engineering, fine-tuning, and post-training of LLMs and other DL models. Delivers 16-24x higher throughput without extra resources.

RapidFire AI 0.16.0

RapidFire Frontend is ready

RapidFire Frontend is ready

Developing with RapidFire AI

MLflow Integration

Full MLflow support for experiment tracking and metrics visualization. A named RapidFire AI experiment corresponds to an MLflow experiment for comprehensive governance

Components

Troubleshooting

For a quick system diagnostics report (Python env, relevant packages, GPU/CUDA, and key environment variables), run:

rapidfireai doctor

If you encounter port conflicts, you can kill existing processes:

lsof -t -i:8850 | xargs kill -9  # jupyter server
lsof -t -i:8851 | xargs kill -9  # dispatcher
lsof -t -i:8852 | xargs kill -9  # mlflow
lsof -t -i:8853 | xargs kill -9  # frontend server
lsof -t -i:8855 | xargs kill -9  # ray dashboard