license: apache-2.0 | |
# Long-VITA-16K | |
Github: https://github.com/VITA-MLLM/Long-VITA | |
- This is the converted weight from https://huggingface.co./VITA-MLLM/Long-VITA-1M. | |
- The original weight is trained on Ascend NPU with MindSpeed. | |
- This weight supports inference and evaluation with Megatron and Transformer Engine on Nvidia GPUs. |