Function _transpose_first_dim

codegeex/megatron/checkpointing.py:205–246

(t, num_splits, num_splits_first, model)

def _transpose_first_dim(t, num_splits, num_splits_first, model):
    input_shape = t.size()
    # We use a self_attention module but the values extracted aren't
    # specific to self attention so should work for cross attention as well
    while hasattr(model, "module"):
        model = model.module
    attention_module = model.language_model.encoder.layers[0].self_attention
    hidden_size_per_attention_head = attention_module.hidden_size_per_attention_head
    num_attention_heads_per_partition = (
        attention_module.num_attention_heads_per_partition
    )
    if num_splits_first:
        """[num_splits * np * hn, h]
        -->(view) [num_splits, np, hn, h]
        -->(tranpose) [np, num_splits, hn, h]
        -->(view) [np * num_splits * hn, h]"""

        intermediate_shape = (
            num_splits,
            num_attention_heads_per_partition,
            hidden_size_per_attention_head,
        ) + input_shape[1:]

        t = t.view(*intermediate_shape)
        t = t.transpose(0, 1).contiguous()
    else:
        """[np * hn * num_splits, h]
        -->(view) [np, hn, num_splits, h]
        -->(tranpose) [np, num_splits, hn, h]
        -->(view) [np * num_splits * hn, h]"""

        intermediate_shape = (
            num_attention_heads_per_partition,
            hidden_size_per_attention_head,
            num_splits,
        ) + input_shape[1:]

        t = t.view(*intermediate_shape)
        t = t.transpose(1, 2).contiguous()
    t = t.view(*input_shape)

    return t


def fix_query_key_value_ordering(model, checkpoint_version):