API

• /model/

  • modeling_glm.py

    • class EncoderDecoder(nn.Module):

• /mpu/ 并行处理单元

  • __init__.py

  • transformer.py

    • class PositionalEmbedding(torch.nn.Module):

    • class ParallelCrossAttention(torch.nn.Module):

    • class ParallelSelfAttention(torch.nn.Module):

    • class ParallelMLP(torch.nn.Module):

    • class ParallelDecoderLayer(torch.nn.Module):

    • class ParallelTransformerLayer(torch.nn.Module):

    • class GPT2ParallelTransformer(torch.nn.Module):