Fastino

Tinker is a training API designed for researchers and developers that allows full control over model fine-tuning while abstracting away the infrastructure complexity. It supports primitives and enables users to build custom training loops, supervision logic, and reinforcement learning flows. It currently supports LoRA fine-tuning on open-weight models across both LLama and Qwen families, ranging from small models to large mixture-of-experts architectures. Users write Python code to handle data, loss functions, and algorithmic logic; Tinker handles scheduling, resource allocation, distributed training, and failure recovery behind the scenes. The service lets users download model weights at different checkpoints and doesn’t force them to manage the compute environment. Tinker is delivered as a managed offering; training jobs run on Thinking Machines’ internal GPU infrastructure, freeing users from cluster orchestration.

Tülu 3 is an advanced instruction-following language model developed by the Allen Institute for AI (Ai2), designed to enhance capabilities in areas such as knowledge, reasoning, mathematics, coding, and safety. Built upon the Llama 3 Base, Tülu 3 employs a comprehensive four-stage post-training process: meticulous prompt curation and synthesis, supervised fine-tuning on a diverse set of prompts and completions, preference tuning using both off- and on-policy data, and a novel reinforcement learning approach to bolster specific skills with verifiable rewards. This open-source model distinguishes itself by providing full transparency, including access to training data, code, and evaluation tools, thereby closing the performance gap between open and proprietary fine-tuning methods. Evaluations indicate that Tülu 3 outperforms other open-weight models of similar size, such as Llama 3.1-Instruct and Qwen2.5-Instruct, across various benchmarks.