Model weights and dataset for CSLM (Efficient Training for Cross-lingual Speech Language Models)