> ## Documentation Index
> Fetch the complete documentation index at: https://docs.blindference.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Installation

> All ways to install the Blindference node runtime

# Installation

## Option 1: PyPI (Recommended)

```bash theme={null}
pip install blindference-node
```

For GPU-accelerated local inference:

```bash theme={null}
pip install "blindference-node[gpu]"
```

For development (tests, linting):

```bash theme={null}
pip install "blindference-node[dev]"
```

## Option 2: From Source

```bash theme={null}
git clone https://github.com/baync180705/Blindference-node.git
cd Blindference-node
pip install -e .
```

For GPU support from source:

```bash theme={null}
pip install -e ".[gpu]"
```

## Option 3: Docker

```bash theme={null}
docker build -t blindference-node .
docker run -e BLF_KEY_PASSWORD=secure_password \
  -e BLF_ICL_ENDPOINT=https://icl.blindference.xyz \
  blindference-node run
```

Or use docker-compose:

```bash theme={null}
docker-compose up
```

## System Requirements

### Minimum (Mock Inference)

| Resource | Requirement    |
| -------- | -------------- |
| Python   | 3.10+          |
| CPU      | 1 core         |
| RAM      | 512MB          |
| Network  | Outbound HTTPS |
| GPU      | None           |

### Recommended (Production)

| Resource | Requirement               |
| -------- | ------------------------- |
| Python   | 3.11+                     |
| CPU      | 2+ cores                  |
| RAM      | 2GB                       |
| Network  | Stable internet           |
| TPM      | TPM 2.0 chip (for tier 1) |

### Optional (Local Model Inference)

| Resource | Requirement      |
| -------- | ---------------- |
| GPU      | NVIDIA with CUDA |
| VRAM     | 4GB+             |
| vLLM     | Installed        |

## Verify Installation

```bash theme={null}
blindference-node --version
# Expected: blindference-node, version 0.3.0
```

## Troubleshooting Installation

### Issue: `pip install` fails with compilation errors

**Cause**: Missing system dependencies for `cryptography` or `web3`.

**Fix** (Ubuntu/Debian):

```bash theme={null}
sudo apt-get update
sudo apt-get install -y python3-dev libssl-dev libffi-dev build-essential
```

**Fix** (macOS):

```bash theme={null}
brew install openssl libffi
```

### Issue: `blindference-node: command not found`

**Cause**: pip installed to a location not in your PATH.

**Fix**:

```bash theme={null}
# Find the binary
python -m site --user-base
# Add to PATH:
export PATH="$PATH:$(python -m site --user-base)/bin"
```

Or use:

```bash theme={null}
python -m blindference_node.cli --help
```

### Issue: GPU inference fails with `CUDA not available`

**Cause**: vLLM requires CUDA but GPU drivers are missing.

**Fix**:

```bash theme={null}
# Check CUDA availability
python -c "import torch; print(torch.cuda.is_available())"

# If False, install NVIDIA drivers and CUDA toolkit
# Or run without GPU:
BLF_SUPPORTED_MODELS=qwen2.5-7b blindference-node init
```