Fine-tuning on consumer hardware. Memory-efficient training. Layer streaming. Open source ML infrastructure.
Fine-tune large models on small GPUs