Unsloth Docs | Unsloth Documentation
Train your own model with Unsloth, an open-source framework for LLM fine-tuning and reinforcement learning.
Summary
Tinker is a training API for fine-tuning large language models from Thinking Machines Lab, supporting supervised fine-tuning, reinforcement learning, LoRA, and vision-language models.
It offers both a high-level Cookbook with automatic training loops and a low-level API for manual control, with reference guides for setup, SFT, RL, loss functions, rendering, models/LoRA, and recipes. Installation uses pip packages `tinker` and `tinker-cookbook`, and example code demonstrates creating a LoRA training client, preparing tokenized data, running forward/backward + optimizer steps, and sampling from the trained model.