Results for “stm32”

Launch distributed Megatron-LM training jobs on a SLURM cluster with a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules, container conventions, monitoring, and per-rank failure diagnosis.

Train and run Mamba state-space models with O(n) complexity, achieving faster inference and longer context than Transformers.

Train language models with SimPO, a reference-free preference optimization method that outperforms DPO without needing a reference model.

Trains LLMs with SimPO, a reference-free preference optimization method that outperforms DPO, using configurable hyperparameters and workflows for various models and tasks.