AI Tools / AI products & services
Marin is an open-source framework, research program, and community for the research and development of foundation models, with a focus on large language models. It supports data curation, transformation, filtering, tokenization, pretraining, posttraining, and evaluation while recording processes, experiments, decisions, and failed runs. Experiments are defined as dependent steps and can run from CPU configurations to large GPU or TPU clusters.
An open-source framework for reproducible foundation-model research, covering data curation, filtering, tokenization, training, and evaluation. It records each step, including failed experiments, and scales from CPU models to large TPU deployments.