OpenSML updates https://www.open-sml.com/ New models, research and updates from OpenSML. en Making training faster without making the model smaller https://www.open-sml.com/research/making-training-faster/ https://www.open-sml.com/research/making-training-faster/ Tue, 06 Oct 2026 00:00:00 GMT Sequential accumulation reduced peak active memory from 87.4 to 26.8 GiB on one worker. Workload and cache tuning then delivered roughly 27% more throughput in short pilots. Distributed training. Shared progress. https://www.open-sml.com/research/distributed-training/ https://www.open-sml.com/research/distributed-training/ How parallel computation, workload balance and gradient synchronization bring many workers into one training process. Introducing OpenSML-150M https://www.open-sml.com/models/opensml-150m/ https://www.open-sml.com/models/opensml-150m/ Explore the OpenSML-150M release, model weights, benchmarks and the research behind it.