sub_modules_to_wrap:
  - megatron.core.transformer.transformer_layer.TransformerLayer
  - megatron.core.models.common.embeddings.language_model_embedding.LanguageModelEmbedding
  - megatron.core.models.common.embeddings.rotary_pos_embedding.RotaryEmbedding
param_dtype: "bf16"
reduce_dtype: "fp32"
cast_forward_inputs: True
ignored_modules:
  - mindspeed_mm.models.vision.vision_model.VisionModel
recompute_modules:
  - megatron.core.transformer.transformer_layer.TransformerLayer