Skip to main content

Speedtronic

GPU-agnostic PyTorch training, built for speed.

Speedtronic 2.0.0 is an alpha, architecture-neutral Python training framework. A run is assembled from a validated configuration, a registered torch.nn.Module, a data source, AdamW or hybrid Muon, an optional scheduler, precision handling, checkpoints, metrics, and—optionally—a Hub-backed DumbDiLoCo coordinator.

Documentation scope

This site documents every local implementation module, every top-level class and function, every method, the v1 and v2 repository tests, both shipped configurations, both examples, and the packaging surface. The generated source inventory is rebuilt from the Python AST during the documentation build. Documentation coverage is not the same as runtime test coverage; see the test map.

Capability map​

AreaWhat Speedtronic 2.0.0 provides
DevicesCPU, CUDA, and Apple MPS selection
Model contractAny module following Speedtronic's loss/batch protocol
Reference modelDecoder-only transformer with RoPE, GQA, SwiGLU, RMSNorm, and tied embeddings
OptimizationAdamW, hybrid Muon/Muon+, cautious updates, gradient accumulation, clipping, warmup, cosine or constant schedule
PrecisionFP32, FP16, BF16, CUDA GradScaler, fused-AdamW probing
PerformanceOptional torch.compile, opt-in CUDA stream backprop, gradient checkpointing, worker prefetch, pinned memory
DataDeterministic synthetic data, streamed UTF-8 text, serialized datasets, programmatic datasets
PersistenceAtomic local checkpoints with model/optimizer/scheduler/RNG/counters
DistributedAsynchronous DumbDiLoCo over a Hugging Face model repository
ObservabilityText/JSONL metrics plus callable, W&B, and TensorBoard hook adapters

One-minute path​

  1. Install and run the CPU smoke test.
  2. Read core concepts.
  3. Choose a tutorial: custom model, custom data, performance, checkpointing, or observability.
  4. Use the configuration reference and complete source inventory while integrating.

System shape​

Documentation map​

Learn the system​

  • Architecture explains composition, training order, and state ownership.
  • Runtime and Trainer documents the public training API and its contracts.
  • Configuration lists every field, default, normalization rule, and precedence edge case.

Explore v2​

Build an integration​

Operate the framework​

Project identity​

  • Package/distribution name: speedtronic
  • Version: 2.0.0
  • Python: >=3.10
  • PyTorch: >=2.1
  • License: Apache-2.0
  • Default training device: auto → CUDA, then MPS, then CPU
  • Default checkpoint directory: <run.output_dir>/checkpoints
  • Distributed transport: Hugging Face model repository files