Glossary
Unsloth
An open fine-tuning library that uses hand-written Triton kernels and a manual gradient implementation to run LoRA and QLoRA fine-tuning roughly 2x faster than the Hugging Face baseline.
A fast fine-tuningtrainingContinued training of a pretrained base model on a smaller, task-specific dataset to specialize its behavior without retraining from scratch. Open full entry library that targets consumer-GPUsiliconA massively parallel processor originally designed for graphics, repurposed since the 2010s as the dominant compute substrate for both training and inference of large neural networks. Open full entry LoRAtrainingA parameter-efficient fine-tuning method that injects small low-rank adapter matrices into a frozen base model, training a tiny fraction of weights instead of the full model. Open full entry and QLoRAtrainingA fine-tuning method that combines 4-bit quantization of the frozen base model with LoRA adapters, making large-model fine-tuning fit on a single consumer GPU. Open full entry workflows. Unsloth replaces several standard PyTorch operations in the training path with hand-written Triton kernels and a manually- implemented backward pass, giving ~2x speedup on supported models and slightly reduced memory.
The use case is single-GPUsiliconA massively parallel processor originally designed for graphics, repurposed since the 2010s as the dominant compute substrate for both training and inference of large neural networks. Open full entry fine-tuningtrainingContinued training of a pretrained base model on a smaller, task-specific dataset to specialize its behavior without retraining from scratch. Open full entry on consumer cards (RTX 4090, 4080, A6000). Users can fit larger context windows or bigger LoRAtrainingA parameter-efficient fine-tuning method that injects small low-rank adapter matrices into a frozen base model, training a tiny fraction of weights instead of the full model. Open full entry ranks for the same VRAM budget. The library supports LlamaweightsMeta's open-weight model family, the most widely deployed open release through 2024 to 2026, released under the source-available Community License with an MAU cap and acceptable-use clause. Open full entry , MistralweightsA French open-weight model family from Mistral AI, released mostly under Apache 2.0 with strong performance per parameter and notable MoE variants (Mixtral, Mixtral 8x22B). Open full entry , QwenweightsAlibaba's open-weight model family, leading the multilingual and Chinese-language open-weight space, released under Apache 2.0 with sizes from 0.6B to 235B parameters. Open full entry , GemmaweightsGoogle's open-weight model family derived from Gemini research, with source-available licensing that includes an acceptable-use clause and license-revocation hook. Open full entry , Phi, and other major open weightsweightsA model release that publishes the trained parameters under some downloadable license, distinct from "open source" which (per OSAID) also requires data and training-code openness. Open full entry families.
Apache 2.0governanceA permissive open-source license used by most open-weight model releases (Llama from 4 onward partial, Qwen, Mistral, DeepSeek, Falcon), allowing commercial use without acceptable-use restrictions. Open full entry licensed core; some performance optimizations sit behind a paid Pro tier.
Full coverage at /projects/unsloth.