pytorch-lightning
Overview
PyTorch Lightning is a lightweight wrapper for PyTorch that decouples the research code from the engineering code. It automates 40+ engineering details like epoch loops, optimization, and hardware acceleration, while allowing researchers to retain full control over the model logic.
When to Use
Use Lightning when you want to scale models to multi-GPU or multi-node environments without changing training code, or when you want to eliminate boilerplate for logging, checkpointing, and reproducibility.
Decision Tree
- Are you testing code logic on a small subset?
- YES: Use
Trainer(fast_dev_run=True).
- YES: Use
- Do you need to scale to multiple GPUs?
- YES: Set
accelerator='gpu'anddevices=Nin the Trainer.
- YES: Set
- Do you have logic that is non-essential to the model (e.g., special logging)?
- YES: Implement it as a
Callback.
- YES: Implement it as a
Workflows
More from cuba6112/skillfactory
ollama-rag
Build RAG systems with Ollama local + cloud models. Latest cloud models include DeepSeek-V3.2 (GPT-5 level), Qwen3-Coder-480B (1M context), MiniMax-M2. Use for document Q&A, knowledge bases, and agentic RAG. Covers LangChain, LlamaIndex, ChromaDB, and embedding models.
17unsloth-sft
Supervised fine-tuning using SFTTrainer, instruction formatting, and multi-turn dataset preparation with triggers like sft, instruction tuning, chat templates, sharegpt, alpaca, conversation_extension, and SFTTrainer.
6torchaudio
Audio signal processing library for PyTorch. Covers feature extraction (spectrograms, mel-scale), waveform manipulation, and GPU-accelerated data augmentation techniques. (torchaudio, melscale, spectrogram, pitchshift, specaugment, waveform, resample)
5pytorch-onnx
Exporting PyTorch models to ONNX format for cross-platform deployment. Includes handling dynamic axes, graph optimization in ONNX Runtime, and INT8 model quantization. (onnx, onnxruntime, torch.onnx.export, dynamic_axes, constant-folding, edge-deployment)
5unsloth-lora
Configuring and optimizing 16-bit Low-Rank Adaptation (LoRA) and Rank-Stabilized LoRA (rsLoRA) for efficient LLM fine-tuning using triggers like lora, qlora, rslora, rank selection, lora_alpha, lora_dropout, and target_modules.
4pytorch-quantization
Techniques for model size reduction and inference acceleration using INT8 quantization, including Post-Training Quantization (PTQ) and Quantization Aware Training (QAT). (quantization, int8, qat, fbgemm, qnnpack, ptq, dequantize)
3