ML Frameworks

PyTorch, TensorFlow, JAX, and HuggingFace Transformers — the deep learning runtime infrastructure used for training and fine-tuning models.

← Prev ← DevOps & Infrastructure

Notes

• Deep Learning Frameworks — PyTorch/TF/JAX architecture, core training loop APIs, and deployment options
• Framework Comparison — decision guide: when to use PyTorch vs TF vs JAX, side-by-side training loops
• PyTorch Patterns — custom datasets, AMP, gradient accumulation, torch.compile, gradient checkpointing
• HuggingFace Usage — Pipeline, AutoModel, Trainer API, datasets library, Hub integration

Links

• DevOps & Infrastructure
• Distributed Training
• PEFT and LoRA