Labels

• arch: amd64

  x86-64 specific (AVX2, SIMD, future)
• arch: arm64

  ARM64 specific (Apple Silicon, NEON)
• arch: wasm

  WebAssembly support
• area: api

  Public API surface (tensor/, nn/, optim/)
• area: autodiff

  Automatic differentiation, gradient tape
• area: cli

  CLI tools (born, born-bench, born-convert)
• area: cpu

  CPU backend, element-wise ops, BLAS
• area: examples

  Example projects (MNIST, CNN, GPU)
• area: generate

  Text generation, LLM inference pipeline
• area: gguf

  GGUF loader, K-quant dequantization
• area: nn

  NN modules (Linear, Conv2D, RMSNorm, Embedding)
• area: onnx

  ONNX model import, 30+ operators
• area: optim

  Optimizers (SGD, Adam)
• area: serialization

  Model save/load (.born format), mmap
• area: tensor

  Core tensor ops, shapes, dtypes, Backend interface
• area: tokenizer

  Tokenizers (TikToken, BPE)
• area: webgpu

  WebGPU/GPU backend, WGSL shaders
• bench: vs-burn

  Comparison with Burn (Rust)
• bench: vs-onnxrt

  Comparison with ONNX Runtime
• bench: vs-pytorch

  Comparison with PyTorch
• effort: 1

  Trivial, < 1 hour
• effort: 13

  Epic, 2+ weeks (split)
• effort: 2

  Small, 1-4 hours
• effort: 3

  Medium, ~1 day
• effort: 5

  Large, 2-3 days
• effort: 8

  Very large, ~1 week
• good first issue

  Good for newcomers
• help wanted

  Extra attention needed
• mentor available

  Maintainer will mentor
• model: born

  Native .born format